profile
viewpoint
David Scott djs55 Docker Inc Cambridge http://dave.recoil.org/ Work for Docker, enthusiast of Unikernels, OCaml and Mirage hacker.

djs55/blktap 2

blktap, vhd stuff

djs55/CloudFS 1

CloudFS is a prototype distributed VM storage system that allows VMs to run without using a SAN for storage

amirmc/mirage-www 0

Website infrastructure for openmirage.org

djs55/alcotest 0

A lightweight and colourful test framework

djs55/appindicator 0

Unity AppIndicator binding for Go

djs55/astring 0

Alternative String module for OCaml

djs55/backtrace 0

Helper functions to preserve and transport exception backtraces

djs55/book 0

V2 of Real World OCaml

PR opened moby/hyperkit

tty: write filename in a file, don't symlink to /dev

If the osquery tool[1] is running and scanning a directory containing the tty symlink, it will follow the link to /dev/ttys* and start some kind of infinite loop involving the debug console. This causes hyperkit's CPU to spike and for the console itself to be spammed with it's own output.

The easiest workaround is to not create the symlink, but write the name of the tty in a file instead. Therefore for debugging we can still run

screen $(cat ~/Library/Containers/com.docker.docker/Data/vms/0/tty)

(or equivalent)

Reported originally around here[2]

[1] https://osquery.io/downloads/ [2] https://github.com/docker/for-mac/issues/3499#issuecomment-639140844

Signed-off-by: David Scott dave@recoil.org

+6 -4

0 comment

1 changed file

pr created time in 9 hours

create barnchdjs55/hyperkit

branch : osquery

created branch time in 9 hours

push eventdjs55/hyperkit

David Scott

commit sha 67ca664be53a3ae0d7b4e04690af44d5aeba09a6

vmx: move Hypervisor HV_VMX_CAP_* query to per-vCPU setup Signed-off-by: David Scott <dave@recoil.org>

view details

Antoine Martin

commit sha 3b930ddfcfc546740359ff8332f5143c1dfd0e7f

fix: Fix upstream xhyve crash (virtio-net + UEFI) When using virtio-net with an UEFI firmware (-f bootrom), hyperkit crashes on an assertion. See the related bug on xhyve: https://github.com/machyve/xhyve/issues/164 The complete details are not clear to me, but it seems that during boot, the firmware reads on the interface but doesn't provide big enough read buffers. It seems that the space for VLAN headers is missing. This patch drops the related reads, so normal booting can occur. Network boot probably doesn't work, though. Signed-off-by: Antoine Martin <amartin@unowhy.com>

view details

David Scott

commit sha 2ed48672b53a3e9650163c9db897536001305c09

vmx: inherit default capability bits from the hypervisor There have been some observed initialisation failures on new hypervisors / hardware, caused by us not recognising a new hypervisor feature. Previously the code would print an error and exit if it encountered an unknown feature. However the hypervisor is able to provide a safe default, so we should use that instead. Setting the features we want is a VMCS write and querying the default fields is a VMCS read. Therefore this patch - moves the capability probing into the vCPU setup where a VMCS read is possible - initialises the feature bits with a VMCS read (previously these were all 0) - removes the failure case when we encounter an unknown feature bit (since the default will be safe) Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha b1bffeee52b7606b573b6c291df2086ed3f6ca48

Merge pull request #285 from djs55/vmx-inherit-defaults vmx: inherit defaults from hypervisor

view details

David Scott

commit sha 79c6a4d95e3f8a59f774eb66e3ea333a277292c6

Merge pull request #284 from amartinunowhy/fix/xhyve_virtio_net_crash_with_UEFI fix: Fix upstream xhyve crash (virtio-net + UEFI)

view details

David Scott

commit sha 3715de4896be9f37c84dc4a8f407cd60013cf34c

ocaml: update to mirage 4.0 Some of the dependencies have been renamed (e.g. cstruct.lwt to cstruct-lwt). Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 6414a3c33383c011044ba1df8f6fd4965a52118b

ocaml: ensure the notifications fd is set to non-blocking Since some users have observed spinning, it could be that the system thinks the fd should be blocking but it is actually non-blocking and returning immediately. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha e88ddf6badb17c44b357630b5d6511e3511843c1

ocaml: handle unexpected error cases from the notifications pipe The `Lwt_unix.read` is supposed to return 0 for EOF and >0 for number of bytes read. Error results from the syscall are supposed to be handled by the library. However there is evidence that an error `-1` is leaking through, so add some logging and exit the process instead of spinning. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha c00e311fbc9094d8b611c82556500ef11b5d5c62

ocaml: bump to OCaml 4.10.0 Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 0991dc7c24847a33ddc32a5e6a23813246dc839d

ocaml: remove out-of-date repo subset We should be able to rely on the version constraints in the hyperkit.opam instead. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 461b3f97b98a0cdaa4498f2351f67b9896eb343f

Merge pull request #286 from djs55/update-mirage ocaml: update dependencies and harden the notification loop

view details

push time in 14 hours

push eventdjs55/mutagen

Jacob Howard

commit sha e83df7f98021107dfc52c15218fde13ff5ec2507

Bumped Travis CI to Go version 1.13.6. This commit bumps the Go version used on Travis CI (for testing and builds) to Go 1.13.6. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha e7c68fe06ed659727293b1ec30176e86dfc14b08

Added strategic debugging and tracing to watching. We've had a few issues with watching feedback loops causing continuous scanning, and we may have one right now with issue #189. This commit adds some strategic debugging and tracing data that can be used to resolve these types of problems. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 00a61f7594052e5544dd2d1ee5d3418d28afd3f4

Added path information to scan errors. One of the biggest problems we've had with debugging scan issues is figuring out the path causing the issue, because, unlike Transition, Scan is pass/fail and doesn't report problematic paths. With our previous error handling framework, it was tedious to embed this additional information (though we probably still should have done it), but with Go 1.13 it's trivial. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha bf8cdfe079b3c0ac6be6f553e5a46fc8de1617ff

Bumped version to v0.11.0-beta3. This commit bumps the version to v0.11.0-beta3. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha d91b566a011c016278a8d7960147a63cfdf0cb04

Updated container provisioning script to use v0.11.0-beta3. This commit bumps the container provisioning script to use Mutagen v0.11.0-beta3 as its default version. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 563609ac63217d84066c75db240a5a710ef46d00

Added cache for filesystem behaviors determined via probe files. This commit adds a cache for scans to store filesystem behavior information determined using probe files. This cache is necessary to avoid watching/scanning feedback loops caused by probe files. It isn't used in cases where probe files aren't used. A fix for these feedback loops was previously attempted in c4830be, but this didn't take into account the modification time changes that are seen on probe files' parent directories. Instead, we now don't attempt to ignore probe files when watching, but rather rely on the behavior cache to simply avoid creating them a second time, thereby averting a feedback loop. This design does come with the remote possibility that a device ID could be re-used with a different filesystem with different behavior, but that's quite unlikely given POSIX's requirements for device IDs. Caching is also only used in select and uncommon cases (since probe files are rarely used), so the cross-section for this scenario is quite small. There's also the issue that this cache is currently unbounded in size, but this shouldn't be an issue in practice given the finite number of filesystems on a system. Updates #189, #108, and #44. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 3e4f962c4705bb9f0eef099556a4342e935b3fa6

Fixed transition tests to adjust for recent changes. This commit adjusts transition tests to account for the extra parameter now returned by probing functions. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha e1bafd7bdec0a7d5e9d2d22cf7d7b3c092a6984e

Cleaned up comments related to scan's new behavior cache. It felt like the comments surrounding the addition of the new behavior cache used in scans weren't fully adequate to describe the situation, so this commit goes a bit further towards clarifying the need for this cache. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 06141a12b968953b13e5dc40317d52fe2dc41f67

Restored temporary file ignores in watching. This commit restores the ignores for temporary files that were removed after the creation of the scan behavior cache. Temporary files are used for more than just probe files, and even though probe files are the only files likely to cause watching/scanning feedback loops, it makes sense to optimize by ignoring all temporary files and it's also more consistent with other code. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha b89531e6569dcc09c4e6ecdad381d29a6540a3ba

Bumped version to v0.11.0-beta4. This commit bumps the version to v0.11.0-beta4. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 8bbdc3b6b3a59ea4f252bccc7e754aafe225ac99

Updated container provisioning script to use v0.11.0-beta4. This commit bumps the container provisioning script to use Mutagen v0.11.0-beta4 as its default version. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 8355086da5eca413dfe27c79b00d89e0c73e14c4

Updated release support policy. In order to support a faster release cadence, we've changed the support policy for Mutagen versions, at least in this pre-1.0 era. Minor release series will now only be supported for 1 month after the first release in the next minor release series. This will allow for manageable 1 month release cycles without sacrificing stability. This also means that, as of right now, v0.9.x is no longer supported, but this is likely to have no impact since almost all users have migrated to v0.10.x. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 6dda5da9c51f48ea956a131f50182bc44801e327

Tweaked wording in version support section. This commit makes a minor tweak to the wording in the version support section of the README. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 54b77ce7302c4f77d4c35f282c43d100513d92e4

Added note to avoid confusion with another Mutagen project. There's another active project called Mutagen that's been brought to our attention, so this commit adds a note to the README to help avoid confusion. Fixes #126. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha f6ace75becfe518427dd04a9f67c0bbb31ada4eb

Updated gitignore to exclude project lock files. This commit adds project lock files to the list of files ignored by Git. It also makes the comments in the .gitignore file a bit simpler. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 79181694edaf2fd4bdb6f4e5170957a931c6019c

Cleaned up configuration structure comments. This commit replaces a configuration structure comment in commands to be clearer and more concise. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 5d44741f54a3133a8806f15f87ee062cbced898d

Shortened container name for Mutagen development environment. This commit shortens the container name used for the Mutagen development container. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 3403352981ceeb0514248e9fb68c399984cd91b4

Removed spacing typo in comments. Somehow we had a double space in a comment, so this commit removes it. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha cbcbe1dfd43159489a15f86723819b309871b64f

Removed announcement list link in README. We're going to stop sending messages via the Google Groups mailing list and start using Twitter and Spectrum as our primary communication channels. Unfortunately Google Groups isn't developed much anymore and it's a little dated at this point. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

Jacob Howard

commit sha 89db45a739108a1aeac1578f3cda8db329b7ad39

Added example web application setup. This commit adds an example Mutagen setup using a simple but multi-service web application. It also tweaks some of the READMEs for the other example projects. Signed-off-by: Jacob Howard <jacob@havoc.io>

view details

push time in 21 hours

issue commentdocker/for-win

DNS resolution of short names on Windows not working

Sorry for the delay in replying -- unfortunately we don't have an internal repro of this DNS problem so it's hard to work on.

Docker Desktop has a built-in DNS server which uses the host APIs to resolve names. If the DNS name resolves on the host, it ought to resolve in the VM also. The DNS server on the host is supposed to handle the search domains, rather than the resolver in the VM. I think to understand what is happening we need to see the DNS traffic on the VM, and the DNS traffic on the host.

The DNS traffic in the VM can be displayed with

% docker run --privileged --net=host djs55/tcpdump -n -i eth0 port 53 
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type EN10MB (Ethernet), capture size 262144 bytes
19:46:43.942431 IP 192.168.65.3.34256 > 192.168.65.1.53: 52146+ A? www.google.com. (32)
19:46:43.942474 IP 192.168.65.3.34256 > 192.168.65.1.53: 52425+ AAAA? www.google.com. (32)
19:46:43.944601 IP 192.168.65.1.53 > 192.168.65.3.34256: 52146 1/0/0 A 216.58.204.36 (48)

(to capture to a file you can use docker run --privileged --net=host -v C:\Users\someone:/out djs55/tcpdump -n -s 0 -w /out/dns.pcap -i eth0 port 53)

A tool like wireshark can be used to view the DNS traffic on the host.

If someone with a repro could create some packet traces I'd be grateful.

Falco20019

comment created time in 4 days

PR opened docker/docker.github.io

Desktop: update mutagen file sync documentation

<!--Thanks for your contribution. See CONTRIBUTING for this project's contribution guidelines. Remove these comments as you go.

DO NOT edit files and directories listed in _data/not_edited_here.yaml.
These are maintained in upstream repos and changes here will be lost.

Help us merge your changes more quickly by adding details and setting metadata
(such as labels, milestones, and reviewers) over at the right-hand side.-->

Proposed changes

<!--Tell us what you did and why-->

Recent edge releases of Docker Desktop have updated the mutagen support. This PR re-synchronises the docs with the code.

The changes are:

  • terminology: we use "synchronize" rather than "cache"
  • it's possible to configure it via :delegated
  • it's possible to avoid it via :cached
  • scripting should work better with docker run
  • it's possible to use the global mutagen config file to completely exclude directories from the sync (in both directions, not just one direction)

Unreleased project version (optional)

<!--If this change only applies to an unreleased version of a project, note that here and base your work on the vnext- branch for your project. If this doesn't apply to this PR, you can remove this whole section. Set a milestone if appropriate. -->

This is all released in Edge 2.3.0.2

+75 -7

0 comment

1 changed file

pr created time in 5 days

create barnchdjs55/docker.github.io

branch : update-mutagen

created branch time in 5 days

push eventdjs55/docker.github.io

Ali Oğuzhan Yıldız

commit sha 9031360ae77681cd8e12e0eae25f2ffdb8ed66af

rename psycopg2 to psycopg2-binary

view details

Ali Oğuzhan Yıldız

commit sha f2349251f10bed99a9228a151875409da83d4dec

Merge branch 'master' into patch-1

view details

Erik Martin-Dorel

commit sha c4b7cf3f5b1f06e0268b6036ef2ab3a98a2faa57

docs(install/linux/docker-ce/debian): Buster 10 = stable

view details

Erik Martin-Dorel

commit sha 651ae2d507fac4fa5cad983cdbc0288158164965

docs(install/linux/docker-ce/debian): s/gnupg2/gnupg-agent/ * This patch makes both Debian [1] and Ubuntu [2] install instructions identical (and still compatible with the supported versions). [1] https://docs.docker.com/install/linux/docker-ce/debian/ [2] https://docs.docker.com/install/linux/docker-ce/ubuntu/ * Moreover, the gnupg2 package [3] (unlike gnupg-agent [4]) is now marked as dummy transitional package in Debian Stretch 9: [3] https://packages.debian.org/stretch/gnupg2 (dummy transitional package) [4] https://packages.debian.org/stretch/gnupg-agent

view details

Lee Van Dyke

commit sha ddbfc9f8ea49ea7955caa5c2f3c8687d737ddf29

Update fedora.md Current instructions don't work, have to either reboot or maybe logout for Cgroups mod.

view details

Isaiah Baker

commit sha 30b503904e509218485b013d2c6e3989d9b1cc20

Remove `restart` and `stop` from command list. UCP commands `restart` and `stop` have been deprecated and are no longer available in UCP 3.2. https://docs.docker.com/ee/ucp/release-notes/#deprecations

view details

Sebastiaan van Stijn

commit sha 2f99b92dcc53c57fbeab0fac22f10228d38cf512

Merge pull request #10513 from Zayah117/patch-1 Remove `restart` and `stop` from command list.

view details

hhchou

commit sha d90d0a47c78fd7058fda32d2e154a84375fc9254

Update part2.md (#10457) Updated the dockerfile comments consistent with ending period marks.

view details

Sebastiaan van Stijn

commit sha 36a0ea227c48dedf3f200e1ceda113156e0ec52d

Merge pull request #10397 from erikmd/update-install-debian Update install/linux/docker-ce/debian.md

view details

jureaky

commit sha cdee2dd67c2be064c9ff2a7fc31364cea601e4ef

Fix wrong device path in NFSv4 volume example (#10429) In section Create a service which creates an NFS volume, NFSv3 command states right device path(:/var/docker-nfs), but NFSv4 command is not configured to refer that path. So updated the path.

view details

Sebastiaan van Stijn

commit sha 331554f4b3eece9f76e16592a2eb43213c4fad95

Fix various links that were generating URLs with `.md` (#10548) * Fix incorrect links in compose section there's a bug causing wrapped links to not work, and replacing some links to point to the .md file, so that IDE's can check if the anchors are valid. Also replaced some links to point to their new location. Signed-off-by: Sebastiaan van Stijn <github@gone.nl> * engine/swarm: update links Signed-off-by: Sebastiaan van Stijn <github@gone.nl> * Fix various broken links There's a bug in the "jekyll-relative-links" plugin that causes wrapped links to not work. Also replacing some links to point to the .md file, so that IDE's can check if the anchors are valid. Finally, replaced some links to point to their new locations, so that users don't get redirected.. Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Sebastiaan van Stijn

commit sha 9a8b866072edb830ff439bf9297e138c3532518b

remove Jenkinsfile from dockerignore There's no longer a Jenkinsfile in this repo, so we can remove it Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Usha Mandya

commit sha 7500f47f4188e2b202870e1322b8100a9fe273d7

Merge pull request #10553 from thaJeztah/update_ignores remove Jenkinsfile from dockerignore

view details

Usha Mandya

commit sha 4ba1276f4c8c63ed82f2aeacd28da183e750f884

WSL 2 backend updates for Edge release

view details

Usha Mandya

commit sha c34ceb9c27b2fe062e277da791b6e11903c8989d

Move networking topics Moved the Networking topics to the Run your app in production node as they logically belong to this section.

view details

Usha Mandya

commit sha 83768da29a2fc332d51ae1040da03ee8ba3b62d6

Move Storage topics Moved the 'Manage application data' topics to 'Run your app in production' section in the TOC.

view details

Mathieu Champlon

commit sha 7786c0bf30637473530241236f16a4970713dbed

Add Docker Desktop Stable 2.2.0.5 release notes

view details

Usha Mandya

commit sha 9b90d98c8e5ca458bd8014879a4b7d3588540754

Merge pull request #10559 from usha-mandya/engdocs-347 Move Networking and Storage topics

view details

Usha Mandya

commit sha 199907a995a32c5947d73056eb190cc36b3af826

Merge pull request #10527 from mat007/desktop-stable-2.2.0.5-release-notes Add Docker Desktop Stable 2.2.0.5 release notes

view details

Usha Mandya

commit sha 51898b4920e120b1b0225991986edf5327867f1d

Merge pull request #10543 from usha-mandya/desktop-2551 WSL 2 backend updates for Edge release

view details

push time in 5 days

Pull request review commentdocker/docker.github.io

RFC: Improve Desktop filesharing docs

 You can also move the disk image to a different location. If you attempt to move  #### File sharing -Choose the local directories you'd like to share with your containers. File sharing is required for volume mounting if the project lives outside of the `/Users` directory. In that case, share the drive where the Dockerfile and volume are located. Otherwise, you get `file not found` or `cannot start service` errors at runtime`.+Use File sharing to allow local directories on the Mac to be shared with Linux containers.+This is especially useful for+editing source code in an IDE on the host while running and testing the code in a container.+By default the `/Users` directory is shared. If your project is outside this directory then it must be added

This PR no longer includes the Mac stuff.

djs55

comment created time in 5 days

push eventdjs55/docker.github.io

Ali Oğuzhan Yıldız

commit sha 9031360ae77681cd8e12e0eae25f2ffdb8ed66af

rename psycopg2 to psycopg2-binary

view details

Ali Oğuzhan Yıldız

commit sha f2349251f10bed99a9228a151875409da83d4dec

Merge branch 'master' into patch-1

view details

Beni Rupp

commit sha 4d5c70525e0b55ccf9a8664959aa7b439d2805b0

Add description for unless-stopped restart policy

view details

David Scott

commit sha ea743e0e6bc25d796ba93ead247a0ce0d7c5644d

Mac: add some examples and best-practices for mutagen A future Docker Desktop Edge will bundle mutagen. Signed-off-by: David Scott <dave.scott@docker.com>

view details

Sebastiaan van Stijn

commit sha 987312af65aa4aee57b5c0cbbbf804cd171159ee

Move Google Analytics to include, and prepare for making it optional Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Usha Mandya

commit sha e18959db805c91a5f2e29ca21896344ceaabfbb6

Merge pull request #10848 from thaJeztah/conditional_analytics Move Google Analytics to include, and prepare for making it optional

view details

Sebastiaan van Stijn

commit sha a7806de7c56672370ec17c35cf9811f61a800a42

Desktop: fix some broken links/images and minor markdown-touch up Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Usha Mandya

commit sha 4c9701c26a82253fd20f917784ee4ec644895135

Merge pull request #10849 from thaJeztah/broken_images Desktop: fix some broken links/images and minor markdown-touch up

view details

Usha Mandya

commit sha c60f87dd601b73b97a02c8b9053e0aad85a77741

Updates to mutagen doc

view details

Sebastiaan van Stijn

commit sha e9547a8397a0e898fe3149ebe0e16b20a45f3959

scss: various small fixes and linting issues - Add generic font classes where missing (except for FontAwesome cfonts) - Combined some rules with their shorthand form - Removed unneeded units (0px -> 0) - Fix some invalid values - Commented out empty styles in _github.scss Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Sebastiaan van Stijn

commit sha 9d446cfd442383d807204118ebf8bd87347b7ed3

Landing page: apply icons using CSS, and remove duplication This patch: - removes the icon images from the landing page, instead setting them as a background-image through css. - replaced the "community-resources" and "play-with-docker" styles in favor of a conditional style (based on cards linking to external websites) - changes the minimal height of cards for mobile devices (there was a big amount of padding on the smallest size, which didn't add much value). Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Sebastiaan van Stijn

commit sha ba7413baffb73d8f04a1d38e971d5349bc07db5a

Landing page: left-align icon on small screen sizes Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Sebastiaan van Stijn

commit sha fed0723f2847cdc5f8bbc05f57733599204bfbff

Landing page: reduce tab-width to prevent wrapping The tabs on the "help by project" card were wrapping on iPhone, so reducing the padding to prevent this. Signed-off-by: Sebastiaan van Stijn <github@gone.nl>

view details

Usha Mandya

commit sha d72ba2f5e541d767e6da22b5e692549e3186c3a6

Merge pull request #10850 from usha-mandya/djs55/mutagen Update Mutagen doc

view details

Usha Mandya

commit sha 8716eb0b98a9a82bec6a3c14f7f3f4cc905cbc29

Add release notes for Desktop Edge 2.3.1.0 (#10851) * Add release notes for Edge 2.3.1.0 * Add relase date and download URL

view details

Usha Mandya

commit sha ba08845b64d6b9f4387148ab878b1e7dafaaf50f

Improvise get started docs

view details

Scott Fredericksen

commit sha fca82492f429b98acae2021cd48b5fba12377718

Update install.md The install instructions are missing a critical step that requires the user to be added to the "docker-users" group. Without this, the user will be prevented from running Docker containers.

view details

Jan Kanis

commit sha b4ab954d3ffc18f59569afff851f8cbad6a5d8a1

fix description for ptrace in seccomp.md The existing description leads the reader to believe that dropping CAP_SYS_PTRACE already blocks all ptraces. That is not true, it only blocks ptracing arbitrary processes. Ptracing child processes is still allowed.

view details

Stefan Scherer

commit sha d1c269e58d74bf34823e2fc1cafcda6d7720b867

Cleanup WSL tech preview links Signed-off-by: Stefan Scherer <stefan.scherer@docker.com>

view details

Sebastiaan van Stijn

commit sha 0f4122dec2971cb65da951f595857806f94f8ee7

Merge pull request #10883 from StefanScherer/fix-wsl-preview-links Cleanup WSL tech preview links

view details

push time in 5 days

delete branch djs55/hyperkit

delete branch : update-mirage

delete time in 5 days

push eventmoby/hyperkit

David Scott

commit sha 3715de4896be9f37c84dc4a8f407cd60013cf34c

ocaml: update to mirage 4.0 Some of the dependencies have been renamed (e.g. cstruct.lwt to cstruct-lwt). Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 6414a3c33383c011044ba1df8f6fd4965a52118b

ocaml: ensure the notifications fd is set to non-blocking Since some users have observed spinning, it could be that the system thinks the fd should be blocking but it is actually non-blocking and returning immediately. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha e88ddf6badb17c44b357630b5d6511e3511843c1

ocaml: handle unexpected error cases from the notifications pipe The `Lwt_unix.read` is supposed to return 0 for EOF and >0 for number of bytes read. Error results from the syscall are supposed to be handled by the library. However there is evidence that an error `-1` is leaking through, so add some logging and exit the process instead of spinning. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha c00e311fbc9094d8b611c82556500ef11b5d5c62

ocaml: bump to OCaml 4.10.0 Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 0991dc7c24847a33ddc32a5e6a23813246dc839d

ocaml: remove out-of-date repo subset We should be able to rely on the version constraints in the hyperkit.opam instead. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 461b3f97b98a0cdaa4498f2351f67b9896eb343f

Merge pull request #286 from djs55/update-mirage ocaml: update dependencies and harden the notification loop

view details

push time in 5 days

PR merged moby/hyperkit

ocaml: update dependencies and harden the notification loop
  • update to opam package manager version 2
  • refresh all the dependencies

There is some evidence that the notification loop is spinning, due to an Lwt_unix.read call returning -1: https://github.com/docker/for-mac/issues/3499#issuecomment-623960890 . This is not supposed to happen but perhaps it is possible if the fd is set to the wrong blocking mode, such that -1, EAGAIN are returned but not handled correctly.

This PR

  • ensures the non-blocking mode is set correctly
  • logs and exits if the call returns <0, which will highlight the problem and prevent spinning
+74 -5856

2 comments

390 changed files

djs55

pr closed time in 5 days

issue commentdocker/roadmap

[Docker Desktop] Investigate Mac CPU performance

@PhilETaylor thanks for the top screenshot. With no containers running it looks like the processes inside the VM aren't busy -- the Linux VM reports 98% idle. I think this suggests a problem with the hypervisor process itself. Could you try running sudo dtruss -p <hyperkit pid> to see what system calls it is making? (From this comment https://github.com/docker/for-mac/issues/3499#issuecomment-628880539 ). If there are a lot of read calls returning -1 then the experimental hyperkit here might fix it: https://github.com/docker/for-mac/issues/3499#issuecomment-640113288

Are you running a tool called osquery? There seems to be a clash between Docker and it, see https://github.com/docker/for-mac/issues/3499#issuecomment-639140844

nebuk89

comment created time in 5 days

issue commentdocker/roadmap

[Docker Desktop] Investigate Mac CPU performance

@dietervdw-spotify I think there are several different performance issues which look similar.

  • if the CPU is used by com.docker.hyperkit and com.docker.osxfs then the problem probably relates to file sharing. The build in https://github.com/docker/roadmap/issues/12#issuecomment-652894544 attempts to fix this
  • if the CPU is used only by com.docker.hyperkit then there's something else inside the VM using CPU. It's possible to dig in with docker run -it --privileged --pid=host justincormack/nsenter1 and then run top
  • if the CPU is used by the Docker process on the Mac, this is a problem in the UI itself

@PhilETaylor could you log into the VM and see which process (if any) is spinning? Use docker run -it --privileged --pid=host justincormack/nsenter1 and then top

nebuk89

comment created time in 5 days

issue commentdocker/for-mac

High CPU Utilization of Hyperkit in Mac

We've found one problem to do with filesystem event watching and have made a build with a prototype fix. Details are on the roadmap ticket: https://github.com/docker/roadmap/issues/12#issuecomment-652894544

taufek

comment created time in 5 days

issue commentdocker/for-mac

com.docker.hyperkit 100% cpu usage is back again

We've found one problem to do with filesystem event watching and have made a build with a prototype fix. Details are on the roadmap ticket: https://github.com/docker/roadmap/issues/12#issuecomment-652894544

stephenc

comment created time in 5 days

issue commentdocker/roadmap

[Docker Desktop] Investigate Mac CPU performance

We've managed to repro one source of unexpected high CPU in hyperkit and com.docker.osxfs by

  • checking out a large .js project on the host (https://github.com/microsoft/vscode)
  • sharing the code with a container docker run -v ~/vscode:/anywhere alpine sh
  • on the host: building the project with yarn install

The build generates a lot of file events which are all injected into the VM.

Unfortunately there are so many file events generated, it takes a long time for them to be fully processed. The build finishes quickly but hyperkit and com.docker.osxfs remain busy for a very long time.

We have an experimental prototype build which we believe fixes this particular problem, if you'd like to try it. The build is based on Edge and can be downloaded from https://desktop-stage.docker.com/mac/edge/46421/Docker.dmg . The build does 3 things to improve performance:

  1. events are batched, rather than injected one-by-one
  2. if the event queue starts to become full, it will prioritise cache invalidations and drop inotify event injections (since inotify events are expensive to create)
  3. if more than 1000 events are pending, it will give up, drop all the events and invalidate all caches instead.

If you have a chance to try it, let us know what happens! Note that this is an experimental build and won't work with existing containers -- they will need to be re-created. It might be necessary to reset to factory defaults before and after using this build.

nebuk89

comment created time in 5 days

issue commentdocker/for-mac

File system performance improvements

@rfay thanks for the list of questions/comments, very useful!

Regarding wait-for-sync, do you think it would be sufficient if we could make

#!/bin/sh
touch file-on-host
docker exec container stat file-on-host

and

#!/bin/sh
docker exec container touch file-in-container
stat file-in-container

both work by automatically performing the mutagen flush around the docker exec? We do this for docker run already but I'm not 100% sure if this is possible yet with docker exec but I'd like to investigate it next.

Regarding the ~/.mutagen.yml, I'm also concerned about it since it would live outside the project directory. Mutagen supports per-session excludes as well as global ones -- the main question is how best to expose this to the Docker user. I've got a ticket in the backlog to use the .dockerignore file in the root of the sync as the mutagen exclude -- do you think that would work for you? The rationale would be that files not to be uploaded in the docker build context would also be files not to be synchronised.

I think it should be possible to share a subdirectory with :cached or :consistent. I just need to check that actually works correctly.

Thanks again for all the feedback!

yallop

comment created time in 5 days

issue commentdocker/for-mac

File system performance improvements

Thanks for the ongoing input -- this is very useful!

To summarise the current situation on edge 2.3.2.0 is:

Caching option Behaviour if mutagen enabled on dir Behaviour if mutagen disabled on dir
default use mutagen use osxfs in fully consistent mode
:consistent use mutagen use osxfs in fully consistent mode
:cached use mutagen use osxfs in cached mode
:delegated use mutagen enable mutagen and use it

I'm sympathetic to the request for a completely new :mutagen flag (or even better :ludicriousspeed as suggested by @pirog), but I'm concerned about the incompatibility this would cause with existing docker setups (as described by @Nemo64 above). I'm also sympathetic to the concerns about changing the behaviour of the :delegated flag, even if the changes are considered within-spec. I definitely agree that we need mechanisms to achieve the pre-mutagen behaviour. As a next step I'd like to change the implementation to this:

Caching option Behaviour if mutagen enabled on dir Behaviour if mutagen disabled on dir
default use mutagen use osxfs in fully consistent mode
:consistent use osxfs in fully consistent mode use osxfs in fully consistent mode
:cached use osxfs in cached mode use osxfs in cached mode
:delegated use mutagen enable mutagen and use it

This would allow any docker run or docker-compose volume to get back the old behaviour by using :consistent or :cached (rather than :delegated).

I'd also like to improve the mutagen implementation to make it as much of a seamless upgrade of :delegated in as many cases as possible. The recent edge build automatically calls mutagen flush when a container starts and stops, so it's possible to write scripts like

$ touch file-on-host
$ docker run -v ... stat file-on-host

and

$ docker run -v ... touch file-in-container
$ stat file-in-container

However I suspect we can do better. I'd like to investigate whether we can enhance docker exec to also flush, so it's possible to

$ touch file-on-host
$ docker exec ... stat file-on-host

etc.

However I take the point that mutagen isn't a good fit for every project (no one solution is) and will think & discuss more about the possibility of an opt-in or opt-out to the new behaviour. In the short term my focus will be on making the new behaviour as good as possible though :)

Thanks again for all the input!

yallop

comment created time in 7 days

issue commentdocker/for-mac

Configure mutagen mode volumes using options instead of file sharing

@daveisfera yes, if you docker run -v ~/workspace:/workspace:delegated alpine sh and then inspect the container with

$ docker inspect <name> | jq .[0].HostConfig.Binds
[
  "/var/lib/mutagen/desktop09ac2896f1b0722124996ec19378aab0:/workspace:delegated"
]

the source path(s) will be rewritten to /var/lib/mutagen/... if the two-way sync is enabled properly.

kinghuang

comment created time in 11 days

issue commentdocker/for-mac

Configure mutagen mode volumes using options instead of file sharing

Yes, :delegated in compose should set up the two-way sync too.

kinghuang

comment created time in 11 days

issue commentdocker/for-mac

Configure mutagen mode volumes using options instead of file sharing

@ajardin the PR to enable caching on :delegated was merged and released in Edge 2.3.2.0: https://desktop.docker.com/mac/edge/46268/Docker.dmg -- let me know what you think.

The behaviour is now:

  • docker run -v /foo:/bar:delegated will create the two-way sync the first time it's executed. A notification is displayed to warn the user it might take some time.
  • subsequent docker run -v /foo:/bar will use this sync-- it's not necessary to keep adding :delegated but it's probably advisable

It's still an open question what the best behaviour of a docker run -v /foo:/bar is (i.e. without :delegated). Currently it will use the sync if it exists. Perhaps we should only use the sync if if :delegated is provided? Or possibly bypass the sync if :consistent is provided. Feedback welcome!

kinghuang

comment created time in 12 days

pull request commentmirage/ocaml-vchan

xen: fix the example kernel build

I think you need to re-create the empty /conduit tree and set the permissions using https://github.com/mirage/ocaml-vchan/blob/master/lib_test/mirage/init-xenstore.sh i.e.

xenstore-rm /conduit
xenstore-write /conduit ""
xenstore-chmod /conduit b0
djs55

comment created time in 12 days

issue commentdocker/for-mac

Mutagen sync taking several seconds

I think there's still a problem, in one direction of the two-way sync.

After following the docs to exclude a directory from the sync (by creating a named docker volume) we have:

  • all the files on the host inside the directory tree with synchronisation enabled in the UI -- minus the files excluded in the mutagen.yml -- are copied into the VM
  • when the container is created, the named docker volumes are bind mounted on top. This means that writes within the container to the excluded directories will go to the named docker volume(s) and won't be synchronised back to the host.

So the host will still scan the excluded directories and copy the changes in, even though they can't be seen in the container. Writes inside the container will be ignored properly.

The next edge build will allow the global mutagen.yml to be configured to fully exclude directories from both directions of the two-way sync.

I wonder if we should support the existing .dockerignore file, used by docker build. The .dockerignore file could then be checked into the source code repo, which would be better than an out-of-tree global mutagen.yml. What do you think?

f1sherman

comment created time in 13 days

issue commentdocker/for-mac

Configure mutagen mode volumes using options instead of file sharing

@ajardin yes, in fact I have an active PR to add :delegated. I'll share the build when it's in slightly better shape.

kinghuang

comment created time in 19 days

issue commentdocker/for-mac

File system performance improvements

@rfay thanks for the two examples! It's interesting that they both involve docker exec -- I'll have to investigate whether it's possible for us to add automatic flushing around that call.

An result of the latest build is that docker run -v <path being synced with mutagen>:/anywhere alpine echo hello will perform the flush+wait (twice!). Potentially that could be used, although it would look a little strange in scripts to have so many echo hellos that don't look like they do anything. I'll have to think about it a bit more.

yallop

comment created time in 20 days

pull request commentmirage/ocaml-vchan

xen: fix the example kernel build

vchan-unix now builds on opensuse with the xen-evtchn-unix and xen-gnt-unix pins (see linked PRs)

djs55

comment created time in 20 days

push eventdjs55/ocaml-evtchn

David Scott

commit sha af9a374418700b7a18f3233e03ebc5856a166778

opam: add xen-devel depext for suse Signed-off-by: David Scott <dave@recoil.org>

view details

push time in 20 days

PR opened mirage/ocaml-evtchn

opam: add xen-dev depext for suse

The os-distribution is opensuse-leap but the os-family is suse which sounds future-proof.

Problem spotted by ocaml-ci in https://github.com/mirage/ocaml-vchan/pull/135

Signed-off-by: David Scott dave@recoil.org

+1 -0

0 comment

1 changed file

pr created time in 20 days

PR opened mirage/ocaml-gnt

opam: add xen-devel depext for suse

The os-distribution is opensuse-leap but the os-family is suse which sounds future-proof.

Problem spotted by ocaml-ci in https://github.com/mirage/ocaml-vchan/pull/135

Signed-off-by: David Scott dave@recoil.org

+1 -0

0 comment

1 changed file

pr created time in 20 days

create barnchdjs55/ocaml-gnt

branch : opensuse

created branch time in 20 days

create barnchdjs55/ocaml-evtchn

branch : opensuse

created branch time in 20 days

push eventdjs55/ocaml-evtchn

David Scott

commit sha ebc06eaeb96ca356120dd8b11b8c343a5bc60748

xen-evtchn-unix: add depexts for CentOS, Fedora and Alpine Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha d4d89fc5374a9f872ad6cc33fce4a1fc67340be8

travis: re-enable debian-stable, centos, fedora, alpine Signed-off-by: David Scott <dave@recoil.org>

view details

Dave Scott

commit sha 9b0a6060fa17f99001a36c5aade29f282097e0d7

Merge pull request #34 from djs55/add-depexts Add depexts for CentOS, Fedora and Alpine

view details

Anil Madhavapeddy

commit sha 0e31de5b6cb46c58edd1ba7287dfebb5ac2c1d3e

port to dune, upgrade opam metadata, switch to dune-release

view details

Anil Madhavapeddy

commit sha a56087c3c01e1548f8c40bdafa226a26a2d46d22

remove bigarray/unix from evtchn library noticed by @TheLortex in mirage/mirage#969

view details

Anil Madhavapeddy

commit sha 64efabfc8e6b67972b94ced428dbf72ef80af2c1

add lwt-dllist dep

view details

Anil Madhavapeddy

commit sha ea2e06d2d7dba2b1321b6228d3ee2ab7e2483828

switch to opensuse testing

view details

Anil Madhavapeddy

commit sha 6194e9afbfc9b0e47df811b31999928a9891adfb

adjust build to distros we use

view details

Anil Madhavapeddy

commit sha 93f649dcd574b4b553034204f8d4df88d37678eb

Merge pull request #36 from avsm/to-dune port to dune, upgrade opam metadata, switch to dune-release

view details

Anil Madhavapeddy

commit sha 89f3194d9ba7ef774dd40e491c47ca4fdabdf0d8

release 2.1.0

view details

Anil Madhavapeddy

commit sha 61882decb28a9261e3a0f4beef812d4177617c89

add doc entries to opam

view details

Anil Madhavapeddy

commit sha 04e1957ed9ac902fd39b64aabd3d2780e264e778

fix tests to unix only

view details

Craig Ferguson

commit sha a6b872777996728e39d86d8e379d5f569ae489fb

opam: remove the 'build' directive on dune dependency See https://github.com/ocaml/opam-repository/pull/14266

view details

Thomas Gazagnaire

commit sha 21db83938d3f466debf9eadd6dde7ac1d4fd05e0

Merge pull request #37 from CraigFe/remove-dune-build-directive opam: remove the 'build' directive on dune dependency

view details

push time in 20 days

push eventdjs55/ocaml-gnt

Lucas Pluvinage

commit sha a9d8c6181b1df953fc04f026d47c416f99330d2a

Upgrade from jbuild to dune files.

view details

Lucas Pluvinage

commit sha dd984e5c5fb906b7f8fd20cede71cf8fa27b1489

Remove unix and bigarray dependencies.

view details

Anil Madhavapeddy

commit sha 6550ea7097677e97868728a157a6070f46768be0

Merge pull request #34 from TheLortex/fix-dep Update to dune + fix dependency specification.

view details

Anil Madhavapeddy

commit sha e9ef27eebeba6b326efdbb6228484055a6deeb12

finish port to dune/dune-release and fix warnings

view details

Anil Madhavapeddy

commit sha 8743727bb7250850b4301f92cafcab0c76a4ba56

Merge pull request #35 from avsm/port-to-dune finish port to dune/dune-release and fix warnings

view details

Anil Madhavapeddy

commit sha df63dee600ce5c62c72d7ae0a9912a1d4d99eec4

remove definitely wrong link line for -L/usr/lib/xen-4.2

view details

Anil Madhavapeddy

commit sha bc1c25ffa4133f16164655a7df513daeab57b5bd

Avoid possible overflow in error printing buffer

view details

Anil Madhavapeddy

commit sha 71a40f4c3cb39f6da76f395233abc2caaea34c79

minor ocamldoc fixes

view details

Thomas Leonard

commit sha ee654ca3138b748930d5e60e39bcec88958b833e

Mark all unikernel-only APIs as deprecated These functions fail at runtime if called from Unix. A newer low-level API is provided by mirage-xen.

view details

yomimono

commit sha 99b12b2df7fed091a2fdb27a0d17dbcd72fdb496

Merge pull request #36 from talex5/deprecate-lowlevel Mark all unikernel-only APIs as deprecated

view details

Mindy

commit sha 7291cf785eb03e94252863117b57cdea23370f1d

update CHANGES for version 4.0.0

view details

yomimono

commit sha 2ca261a4a36aac5cf73902d709f3a2686fcdf1fe

Merge pull request #37 from yomimono/release-v4.0.0 update CHANGES for version 4.0.0

view details

Craig Ferguson

commit sha a6cff8545ebb9ea8acb16d5abf68fd17f6a4dfa1

opam: remove the 'build' directive on dune dependency See https://github.com/ocaml/opam-repository/pull/14266

view details

Thomas Gazagnaire

commit sha e3029291d5669ec66d9b4b8033d807983a8286ac

Merge pull request #39 from CraigFe/remove-dune-build-directive opam: remove the 'build' directive on dune dependency

view details

push time in 20 days

issue commentdocker/for-mac

File system performance improvements

@ajardin thanks for the confirmation! @rfay for code running inside the container -- what kind of interface would you prefer? Would it make sense to provide an HTTP endpoint which would either trigger the flush and block or could be polled for updates? Or something else?

yallop

comment created time in 20 days

PR opened mirage/ocaml-vchan

xen: fix the example kernel build

The current released mirage command-line doesn't handle the -f <config_client.ml> file properly, see for example https://github.com/mirage/functoria/pull/195 . This PR works around this by separating the client and server uni kernels into separate directories.

For anyone unlucky enough to try this on Ubuntu 20.04 LTS, they'll find the Xen packages are broken so I've added some notes on manually working around that.

I've added a README.md to make it easier for others to run these things.

+93 -4

0 comment

10 changed files

pr created time in 20 days

create barnchdjs55/ocaml-vchan

branch : fix-lib-test

created branch time in 20 days

push eventdjs55/mirage

push time in 20 days

create barnchdjs55/mirage

branch : 3.7.7-type-pv

created branch time in 20 days

issue commentdocker/for-mac

mutagen request: ability to wait for cache sync to complete

@allenpetersen @fdanielsen @andig : regarding the problem of synchronising files before/after running containers, I've made another prototype build with explicit mutagen flush on docker run. The experimental build is: https://desktop-stage.docker.com/mac/edge/45977/Docker.dmg I've also created an index of these builds in a public gist as it's getting a little bit difficult to keep track of them: https://gist.github.com/djs55/0e8957e1e36431a43cbfd3caaffa9165

Using this new build I can do this reliably:

# create a file in a container and then use it immediately from the host
$ docker run -it -v ~/workspace:/foo -w /foo alpine touch new-file && ls ~/workspace/new-file
/Users/djs/workspace/new-file

# create a file on the host and then immediately use it in a container
$ touch ~/workspace/host-file && docker run -it -v ~/workspace:/foo -w /foo alpine stat host-file
  File: host-file
  Size: 0         	Blocks: 0          IO Block: 4096   regular empty file
Device: fe01h/65025d	Inode: 2752523     Links: 1
Access: (0666/-rw-rw-rw-)  Uid: (  501/ UNKNOWN)   Gid: (   20/ dialout)
Access: 2020-06-16 12:31:11.000000000
Modify: 2020-06-16 12:31:11.000000000
Change: 2020-06-16 12:31:11.000000000

Let me know what you think!

allenpetersen

comment created time in 21 days

issue commentdocker/for-mac

File system performance improvements

@Nemo64 @ajardin @rfay (and probably others who I've missed): regarding the problem of synchronising files before/after running containers, I've made another prototype build with explicit mutagen flush on docker run. The experimental build is: https://desktop-stage.docker.com/mac/edge/45977/Docker.dmg I've also created an index of these builds in a public gist as it's getting a little bit difficult to keep track of them: https://gist.github.com/djs55/0e8957e1e36431a43cbfd3caaffa9165

Using this new build I can do this reliably:

# create a file in a container and then use it immediately from the host
$ docker run -it -v ~/workspace:/foo -w /foo alpine touch new-file && ls ~/workspace/new-file
/Users/djs/workspace/new-file

# create a file on the host and then immediately use it in a container
$ touch ~/workspace/host-file && docker run -it -v ~/workspace:/foo -w /foo alpine stat host-file
  File: host-file
  Size: 0         	Blocks: 0          IO Block: 4096   regular empty file
Device: fe01h/65025d	Inode: 2752523     Links: 1
Access: (0666/-rw-rw-rw-)  Uid: (  501/ UNKNOWN)   Gid: (   20/ dialout)
Access: 2020-06-16 12:31:11.000000000
Modify: 2020-06-16 12:31:11.000000000
Change: 2020-06-16 12:31:11.000000000

Let me know what you think!

yallop

comment created time in 21 days

issue commentdocker/for-mac

File system performance improvements

@dvessel although the latest Edge does ignore ~/.mutagen.yml there are some later development builds which you are welcome to try. I'm keeping a list here: https://gist.github.com/djs55/0e8957e1e36431a43cbfd3caaffa9165 . The latest build is https://desktop-stage.docker.com/mac/edge/45494/Docker.dmg which should support ~/.mutagen.yml and also default to posix-raw for symlinks. Let me know if you get a chance to try those (but note they are development snapshots so may not be as stable as officially released builds)

yallop

comment created time in 22 days

issue commentmoby/moby

DNS Cache flush on host system not reflected in Docker containers.

Thanks for your report. The DNS resolution in Docker Desktop uses the host resolver, via DNSServiceRef. I believe the host has a cache in the mDNSResponder process which can be flushed manually. I don't think there's a cache in the Docker stack itself (although I could be wrong).

You should be able to see the DNS queries and responses from the VM with a command like:

$ docker run -it --net=host djs55/tcpdump -n -i eth0 port 53
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type EN10MB (Ethernet), capture size 262144 bytes
13:14:51.821165 IP 192.168.65.3.51819 > 192.168.65.1.53: 63020+ [1au] A? www.google.com. (55)
13:14:51.822475 IP 192.168.65.1.53 > 192.168.65.3.51819: 63020 1/0/0 A 216.58.204.36 (48)

Could you reproduce the problem and add the DNS requests and responses to the ticket?

illeatmyhat

comment created time in 22 days

PR opened mirage/functoria

If `config_file` is changed, update the dune.config

For example

$ cd mirage-skeleton/tutorial/noop
$ mv config.ml config_test.ml
$ mirage configure -f config_test.ml

Previously I would see:

File "dune.config", line 5, characters 8-14:
5 |   (name config)
            ^^^^^^
Error: Module "Config" doesn't exist.
run ['dune' 'exec' '--root' '/home/djs/djs55/mirage-skeleton/tutorial/noop'
     '--' './config.exe' 'configure' '-f' 'config_test.ml']: exited with 1

The dune.config file has:

;; auto-generated by 'mirage configure -- remove these comments to
;; preserve the file after a `mirage clean`

(executable
  (name config)
  (flags (:standard -warn-error -A))
  (modules config_test)
  (libraries mirage))

I think we need to update the name (and the final .exe) if the config_file is changed.

Signed-off-by: David Scott dave@recoil.org

+4 -3

0 comment

1 changed file

pr created time in 23 days

create barnchdjs55/functoria

branch : config

created branch time in 23 days

push eventdjs55/functoria

Thomas Gazagnaire

commit sha fc6c112e519c6eb5f493b4668f7a4c5a321abec6

Fix example of keys in the README Fixes #114

view details

Thomas Gazagnaire

commit sha a0e13c27598592c2b02752f0791a927587765f48

Merge pull request #144 from samoht/readme Fix example of keys in the README

view details

Olle Jonsson

commit sha 09d64557b51c2b60af4a9d6f46b0d6286499a209

README: Travis badge for master branch [ci skip]

view details

Hannes Mehnert

commit sha d59d9c32a55ba899436a427305ced00f2bd39f87

Merge pull request #146 from olleolleolle/patch-1 README: Travis badge for master branch

view details

Hannes Mehnert

commit sha 3c8b3549cb179172f22d4dbddef023eac62d7a30

unify to Rresult.R.msg and result (functoria is 4.03 only anyways)

view details

Hannes Mehnert

commit sha a135eb4fba533046b44c1e72e81431fd3fd49ab1

test 4.05 and 4.06 as well

view details

Hannes Mehnert

commit sha 8c161c5d8e0321b1d402d58edb152a6b242e8c63

restrict travis to 4.04 and 4.06

view details

Gabriel Radanne

commit sha f9f2bb67cb62c310737ea93a14d221d9ba188cb6

Merge pull request #147 from hannesm/minor Minor cleanups

view details

Hannes Mehnert

commit sha 60697c99a0252715730e5d3891919f9adb663f8f

add 4.05 to test matrix

view details

Mindy Preston

commit sha 67d385beccd06f6cc096d350ddf4d550e29bdebc

Merge pull request #148 from hannesm/minor add 4.05 to test matrix

view details

Thomas Gazagnaire

commit sha 804faab9aaeefc2ace7b263df51481208ca89d17

Add opam2 local dir to gitignore

view details

Thomas Gazagnaire

commit sha c9f4c7e53f015adcae88133f1c84330be056c5d8

Fix serialization of ints

view details

Thomas Gazagnaire

commit sha e8b4579e145ad83b89b2a1a80fc90dbb697c353a

Fix compilation of tests for recent version of jbuilder

view details

Thomas Gazagnaire

commit sha 4f3634be836f63c9068d529405006211a35665c6

Merge pull request #152 from samoht/negative-ints Fix support for negative ints in keys

view details

Drup

commit sha 1861cf9a5f31eab5ebd137ed541052ceaf6f3369

Expand the signature manually.

view details

Hannes Mehnert

commit sha ef5cfd009886e3a30639f57fb292e840d029b453

Merge pull request #153 from Drup/modulecrap Expand the signature Functoria.Key manually.

view details

Hannes Mehnert

commit sha 8042a59266a88164dfbf3997e77c5a6c80d622ca

prepare for release

view details

Hannes Mehnert

commit sha 62ed4eafd196beea935c018ca6bd7873beb59c66

[ci skip] set date and version number

view details

Hannes Mehnert

commit sha 086b354a7e059671f3e737afd2728705314ffdf0

Merge pull request #154 from hannesm/minor prepare for release

view details

Hannes Mehnert

commit sha 0048abf17a73f391448d357a07cebc01d89f176d

functoria-runtime.opam: adjust lower ocaml bound to functoria.opam

view details

push time in 23 days

PR opened mirage/mirage

xen: use `type="pv"` rather than `builder=...`

I notice that current mirage doesn't work on xen 4.10+ (as used in e.g. LTS Ubuntu) because they removed the builder argument from the .xl file and replaced it with type.

See https://wiki.xen.org/wiki/Type_Config_Option

Obviously we're also in the process of refactoring Xen support to use pvh as in #1089

I don't know whether we should change it to type = pv in the meantime. I'm making this PR just in case -- I needed to make this change anyway for my local testing.

Signed-off-by: David Scott dave@recoil.org

+1 -1

0 comment

1 changed file

pr created time in 23 days

create barnchdjs55/mirage

branch : type-pv

created branch time in 23 days

push eventdjs55/mirage

Gabriel Radanne

commit sha dd1685ce48adb93282ab1477c75576004c5a8526

Merge pull request #23 from Drup/fix_describe The describe commands accepts keys again.

view details

Drup

commit sha af155cde319089724852519ddccd5a1e80e0eb76

Add env variable handling.

view details

Drup

commit sha 21560e8a0ec6f2c9471924c67d17b431a8edb651

Add Key.flag and Key.proxy

view details

Drup

commit sha 71efd9d9fac25314d82e1267b8bd1a0d04c545f4

Misc improvements about stages.

view details

Drup

commit sha c9403b95dcc5ac2dd91bae52ffc3d18b7eaf1b25

Pin cmdliner.

view details

Thomas Gazagnaire

commit sha dca31a4fe23bfaaf7a83d5cb4221221e3b7a3d55

Merge pull request #24 from Drup/superkeys Various improvement to keys

view details

Thomas Gazagnaire

commit sha 243c7faf0b803c8c75966d338f866ab9c98bbe42

Turn-on warning as errors to avoid ignoring warnings

view details

Thomas Gazagnaire

commit sha ce931897f20fbe5aa3617d8fc65d8219222717b4

Hack to turn-off warning 32 on compat symbols

view details

Thomas Gazagnaire

commit sha 695495d22e426342c7fde4e11b0b6756b7bcb3b4

Add a --color command-line option Fix #18 Inital code from @dbuenzli

view details

Thomas Gazagnaire

commit sha c2b506ea14f6d4c3e614db5ddc2fad91a0f90dd1

Initialize the format parameters in the help subcommand Related to #18

view details

Thomas Gazagnaire

commit sha c56eaf940252b9dd34dc6ff440eaad64572d60f0

Add the --color option to `<main.native> --help` Related to #18

view details

Thomas Gazagnaire

commit sha 5f5fa43bd48e3d1194e1fee4a900ca178b14d4ba

Cut long lines

view details

Thomas Gazagnaire

commit sha 7426803b0c565a01154a5307882330f0c554e826

Minor style updates - remove space before `:` and `;` - remove ocamldoc comments in ml files - run ocp-indent -i *.ml *.mli to get consistent indentation - cut long lines

view details

Thomas Gazagnaire

commit sha d4d5739f2eec97dfd0cb86f3bb3df1acde6344c0

long lines in _oasis

view details

Thomas Gazagnaire

commit sha 72849482a5d791f93dc788fc7cf4551ec9cd6212

More minor style fixes, but in runtime/ this time

view details

Thomas Gazagnaire

commit sha be1272aa72b731b5f249ab2714ebb628b7b55006

Add missing license headers

view details

Thomas Gazagnaire

commit sha 63b0c12389765bf83ec3a79a3fe7214dcd4968de

Fix doc hyperlinks and long lines

view details

Thomas Gazagnaire

commit sha a038d30067248612f424277c5f1d71efc6082e65

More doc tweaks

view details

Thomas Gazagnaire

commit sha 6f55bd4637ea318ef8309a5d7d718ea2cd6586ca

Add doc&release Makefile runes

view details

Thomas Gazagnaire

commit sha b1b5ea0f79e56f31b8e9ed75fe019696906ad1f1

remove capital letter in project name in _oasis

view details

push time in 24 days

delete branch djs55/opam-repository

delete branch : release-qcow-0.11.0

delete time in 25 days

pull request commentocaml/opam-repository

[new release] qcow and qcow-tool (0.11.0)

Thanks again!

djs55

comment created time in 25 days

delete branch djs55/opam-repository

delete branch : release-mirage-block-ramdisk-0.5

delete time in 25 days

pull request commentocaml/opam-repository

[new release] mirage-block-ramdisk (0.5)

No problem, thanks for the merge!

djs55

comment created time in 25 days

issue commentdocker/for-win

"Function not implemented" when trying to access a file within a container. Works fine from Windows. Need to restart Docker to fix.

@mdecurtins there have been some fixes in later builds, if you'd like to experiment. There's a stable-derived build in https://github.com/docker/for-win/issues/5955#issuecomment-639360958, and a later edge-derived build which catches an additional case of this bug in https://github.com/docker/for-win/issues/5955#issuecomment-641255062 .

bertrandgorge

comment created time in a month

issue commentdocker/for-mac

Mutagen - file sync 'error' and results in high CPU

Thanks (sorry for the typos) -- the lifecycle-server.exe at 40% is where the mutagen server component lives. On the host the com.docker.driver.amd64-linux process is where the mutagen client component lives. Have you noticed whether it is successfully synchronising files, even though it is in the error state? I'm not sure whether it's permanently stuck (because of the lstat problem) or whether the problem report is more of an FYI. Is the high CPU load because the tree has a lot of files and it is kept busy by the log writing? If so it might work better if the logs were written to a docker volume instead?

Wilkolicious

comment created time in a month

issue commentdocker/for-mac

Mutagen - file sync 'error' and results in high CPU

Oh you're right about the sync.json, I was looking in the wrong place -- sorry!

If the machine is still in that state, could you log into the VM with docker run -it --pid=host --privileged justincormack/nsenter and run top? I'd like to check which process is using the CPU. I can see a process list in the diagnostics but I don't think I can tell where the 241.5% CPU went :(

Wilkolicious

comment created time in a month

issue commentdocker/for-mac

Mutagen - file sync 'error' and results in high CPU

Thanks for your report. I believe "Error" means the mutagen sync encountered problems but we don't currently display those in the UI :( Also unfortunately the diagnostics code in that build doesn't include them so I can't see them in the logs.

Try running a command like

curl -X GET --unix-socket ~/Library/Containers/com.docker.docker/Data/docker-api.sock http:/localhost/cache/state

-- this should display a list of problems (if there are any).

Regarding the high CPU usage: which process was using the CPU? Was it the main "Docker" process (with the icon) or one of the helper processes (e.g. com.docker....)

Wilkolicious

comment created time in a month

issue commentdocker/for-win

"Function not implemented" when trying to access a file within a container. Works fine from Windows. Need to restart Docker to fix.

@rfay thanks for the example, it also reproduces for me on that build. I think this instance of the bug is in the handling of the //c/Users path. I've got yet another development build if anyone would like to try it: https://desktop-stage.docker.com/win/edge/45823/Docker%20Desktop%20Installer.exe . When I install this build the command works as expected: (after I'm prompted to allow the directory to be shared)

> docker run -v "//c/Users:/junkmount"  --rm busybox sh -c "df /junkmount"
Filesystem           1K-blocks      Used Available Use% Mounted on
grpcfuse             486931040 119100336 367830704  24% /junkmount

Thanks again!

bertrandgorge

comment created time in a month

startedmutagen-io/mutagen

started time in a month

issue commentdocker/for-mac

com.docker.hyperkit 100% cpu usage is back again

@PSIAlt thanks for the feedback. The observation about osquery is quite interesting and is consistent with what @kelvich said above https://github.com/docker/for-mac/issues/3499#issuecomment-639140844 . This is definitely worth investigating.

stephenc

comment created time in a month

issue commentdocker/for-mac

com.docker.hyperkit 100% cpu usage is back again

@cculianu @b1rdex @PSIAlt @stephenc @kelvich regarding the suspicious -1 return from read, I've made a PR to make the code more defensive and to fix one possible cause: https://github.com/moby/hyperkit/pull/286 In case you want to try it, you can download the artifact from the CI like this:

wget https://757-55985023-gh.circle-artifacts.com/0/hyperkit.sym
chmod +x hyperkit.sym 

# backup the current version
mv /Applications/Docker.app/Contents/Resources/bin/com.docker.hyperkit /Applications/Docker.app/Contents/Resources/bin/com.docker.hyperkit.orig

mv hyperkit.sym /Applications/Docker.app/Contents/Resources/bin/com.docker.hyperkit

and then start/restart Docker Desktop. Let me know if this makes any difference at all.

stephenc

comment created time in a month

pull request commentmoby/hyperkit

ocaml: update dependencies and harden the notification loop

@samoht WDYT of this, in particular https://github.com/moby/hyperkit/pull/286/commits/6414a3c33383c011044ba1df8f6fd4965a52118b and https://github.com/moby/hyperkit/pull/286/commits/e88ddf6badb17c44b357630b5d6511e3511843c1 ?

djs55

comment created time in a month

push eventdjs55/hyperkit

David Scott

commit sha 0991dc7c24847a33ddc32a5e6a23813246dc839d

ocaml: remove out-of-date repo subset We should be able to rely on the version constraints in the hyperkit.opam instead. Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

push eventdjs55/hyperkit

Antoine Martin

commit sha 3b930ddfcfc546740359ff8332f5143c1dfd0e7f

fix: Fix upstream xhyve crash (virtio-net + UEFI) When using virtio-net with an UEFI firmware (-f bootrom), hyperkit crashes on an assertion. See the related bug on xhyve: https://github.com/machyve/xhyve/issues/164 The complete details are not clear to me, but it seems that during boot, the firmware reads on the interface but doesn't provide big enough read buffers. It seems that the space for VLAN headers is missing. This patch drops the related reads, so normal booting can occur. Network boot probably doesn't work, though. Signed-off-by: Antoine Martin <amartin@unowhy.com>

view details

David Scott

commit sha b1bffeee52b7606b573b6c291df2086ed3f6ca48

Merge pull request #285 from djs55/vmx-inherit-defaults vmx: inherit defaults from hypervisor

view details

David Scott

commit sha 79c6a4d95e3f8a59f774eb66e3ea333a277292c6

Merge pull request #284 from amartinunowhy/fix/xhyve_virtio_net_crash_with_UEFI fix: Fix upstream xhyve crash (virtio-net + UEFI)

view details

David Scott

commit sha 3715de4896be9f37c84dc4a8f407cd60013cf34c

ocaml: update to mirage 4.0 Some of the dependencies have been renamed (e.g. cstruct.lwt to cstruct-lwt). Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 6414a3c33383c011044ba1df8f6fd4965a52118b

ocaml: ensure the notifications fd is set to non-blocking Since some users have observed spinning, it could be that the system thinks the fd should be blocking but it is actually non-blocking and returning immediately. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha e88ddf6badb17c44b357630b5d6511e3511843c1

ocaml: handle unexpected error cases from the notifications pipe The `Lwt_unix.read` is supposed to return 0 for EOF and >0 for number of bytes read. Error results from the syscall are supposed to be handled by the library. However there is evidence that an error `-1` is leaking through, so add some logging and exit the process instead of spinning. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha c00e311fbc9094d8b611c82556500ef11b5d5c62

ocaml: bump to OCaml 4.10.0 Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha fd6317789bc45693953bced778eaf4fac5806f5b

ocaml: remove out-of-date repo subset We should be able to rely on the version constraints in the hyperkit.opam instead. Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

PR opened moby/hyperkit

ocaml: update dependencies and harden the notification loop
  • update to opam package manager version 2
  • refresh all the dependencies

There is some evidence that the notification loop is spinning, due to an Lwt_unix.read call returning -1: https://github.com/docker/for-mac/issues/3499#issuecomment-623960890 . This is not supposed to happen but perhaps it is possible if the fd is set to the wrong blocking mode, such that -1, EAGAIN are returned but not handled correctly.

This PR

  • ensures the non-blocking mode is set correctly
  • logs and exits if the call returns <0, which will highlight the problem and prevent spinning
+40 -14

0 comment

6 changed files

pr created time in a month

create barnchdjs55/hyperkit

branch : update-mirage

created branch time in a month

PR opened ocaml/opam-repository

[new release] mirage-block-ramdisk (0.5)

In-memory BLOCK device for MirageOS

  • Project page: <a href="https://github.com/mirage/mirage-block-ramdisk">https://github.com/mirage/mirage-block-ramdisk</a>
  • Documentation: <a href="https://mirage.github.io/mirage-block-ramdisk/">https://mirage.github.io/mirage-block-ramdisk/</a>
CHANGES:
  • update to mirage-block.2.0.0 API (@hannes)
+38 -0

0 comment

1 changed file

pr created time in a month

create barnchdjs55/opam-repository

branch : release-mirage-block-ramdisk-0.5

created branch time in a month

release mirage/mirage-block-ramdisk

0.5

released time in a month

created tagmirage/mirage-block-ramdisk

tag0.5

A simple in-memory BLOCK device

created time in a month

push eventmirage/mirage-block-ramdisk

David Scott

commit sha 867463b2d7efc4f36335e54e1192ce0c50685487

Update mirage-block-ramdisk doc to 0.5.

view details

push time in a month

delete branch mirage/mirage-block-ramdisk

delete branch : release.0.5

delete time in a month

push eventmirage/mirage-block-ramdisk

David Scott

commit sha ed21bf7964f7176e5b87708aa5c06b05a460b944

Prepare to release 0.5 Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 470536a1755cae1853377cf7773b5fa50471a820

Merge pull request #22 from mirage/release.0.5 Prepare to release 0.5

view details

push time in a month

PR merged mirage/mirage-block-ramdisk

Prepare to release 0.5

Signed-off-by: David Scott dave@recoil.org

+3 -0

0 comment

1 changed file

djs55

pr closed time in a month

PR opened mirage/mirage-block-ramdisk

Prepare to release 0.5

Signed-off-by: David Scott dave@recoil.org

+3 -0

0 comment

1 changed file

pr created time in a month

create barnchmirage/mirage-block-ramdisk

branch : release.0.5

created branch time in a month

push eventdjs55/opam-repository

hbr

commit sha 6050fe1d0c0811c8bb37b3638e0e9d122b30790a

Release of Alba 0.4.2 new file: packages/alba/alba.0.4.2/opam

view details

Gabriel Scherer

commit sha 028878d738a43a86c380bb2ececb4c1d559422ef

ocaml-secondary-compiler: --disable-graph-lib (disable Graphics) Fixes #16226

view details

Jun Furuse

commit sha 347745bedd5005808b4055b19dadc3e81f29239e

Fixed ppx_test.1.6.0

view details

Spiros Eliopoulos

commit sha 3a7404c3b531c8b60d7981626d151051f3202dd5

+webmachine.0.6.2

view details

Jun Furuse

commit sha f5a488d6aa5cb88a294fdc0dfc6c569c5800bc64

synopsis fix

view details

Francois Berenger

commit sha 28b54752ca27fddd37c71ea984fb5c2ec3e1c2d4

new file: packages/hts_shrink/hts_shrink.2.1.1

view details

Justin Lubin

commit sha 9037796b3bcb7487242d512e3647bb9adcc6439f

[new release] bark (0.1.4) CHANGES: Fix off-by-one error that affected multi-line comments

view details

Victor Dumitrescu

commit sha bff40b8633ed44672a0d9c3bab895bb062002d16

2 packages from project-everest/hacl-star at 0.1

view details

Victor Dumitrescu

commit sha 9fd9e28da4f7a6274ef8e5f032e00b48684e060a

Merge pull request #9 from victor-dumitrescu/opam-publish-hacl-star-hacl-star-raw.0.1 2 packages from project-everest/hacl-star at 0.1

view details

Davide Gessa (dakk)

commit sha 73ac5abbda1cd7403d70efeb3a4248802aad093d

Package lambda.0.1.7

view details

Jeremie Dimino

commit sha 7cd11890a85f30d57a4f2a752c81fd05377972e4

[new release] ocaml-migrate-parsetree (1.7.2) CHANGES: - Remove toplevel `Option` module accidentally added in 1.7.0

view details

Anil Madhavapeddy

commit sha 841af62a55ae7c7a0d286f72af5129ad373b45e4

merge conf-ncurses.1 and ncurses.1+system It's unnecessary to have the solver for this, when one package can express the right dependencies fine. This in turns helps with static analysis of a packages dependencies when we run opam on a host other than the distribution target.

view details

Anil Madhavapeddy

commit sha 710f129644f44664d0d7f07d054ccfdf473a14be

Merge pull request #16239 from justinlubin/release-bark-0.1.4 [new release] bark (0.1.4)

view details

Kate

commit sha 4b12a93e8d9a6cbb5e4a2fe0cbe70dfddc70c0a9

Merge pull request #16241 from diml/release-ocaml-migrate-parsetree-1.7.2 [new release] ocaml-migrate-parsetree (1.7.2)

view details

Anil Madhavapeddy

commit sha 1f01b41f7e66c88d22ec7eb56cea6e774828c6e3

Merge pull request #16235 from camlspotter/ppx_test.1.6.0 Fixed ppx_test.1.6.0

view details

Kate

commit sha 45c5e0200a94a4a1248fb81c88bd0dc0c4cb587a

Merge pull request #16219 from rgrinberg/release-dune-2.5.1 [new release] dune-build-info, dune, dune-configurator, dune-action-plugin, dune-private-libs and dune-glob (2.5.1)

view details

Kate

commit sha 5259387eb3a57ac077b484893b0d4c422702c40b

Merge pull request #16222 from astrada/gapi-gdfuse Add gapi-ocaml 0.3.19, google-drive-ocamlfuse 0.7.21

view details

Kate

commit sha df7c2c86abc8983bc8271300ae8d545f6c5eee09

Merge pull request #16237 from seliopou/webmachine.0.6.2 +webmachine.0.6.2

view details

Kate

commit sha 2ac07d479ea39bc6e1ef6d3ec407eaa940ba6ca9

lambda.0.1.7: Fix dependency constraint

view details

Kate

commit sha 62aa77b2061e02f53373042fcf667328dc875e92

alba.0.4.2: Fix dependency constraint

view details

push time in a month

PR opened ocaml/opam-repository

[new release] qcow and qcow-tool (0.11.0)

Support for Qcow2 images

  • Project page: <a href="https://github.com/mirage/ocaml-qcow">https://github.com/mirage/ocaml-qcow</a>
  • Documentation: <a href="https://mirage.github.io/ocaml-qcow">https://mirage.github.io/ocaml-qcow</a>
CHANGES:
  • Update the build to use dune (@emillon, mirage/ocaml-qcow#112)
  • Update to Mirage 4.0 interfaces (@djs55, mirage/ocaml-qcow#112)
  • LICENSE.md: add title and copyright year range (@waldyrious, mirage/ocaml-qcow#109)
+118 -0

0 comment

2 changed files

pr created time in a month

create barnchdjs55/opam-repository

branch : release-qcow-0.11.0

created branch time in a month

release mirage/ocaml-qcow

0.11.0

released time in a month

created tagmirage/ocaml-qcow

tag0.11.0

Pure OCaml code for parsing, printing, modifying .qcow format data

created time in a month

push eventmirage/ocaml-qcow

David Scott

commit sha 58fa9c477ad2de4cd53fa4313419136c53abb3fc

Update qcow doc to 0.11.0.

view details

push time in a month

push eventdjs55/ocaml-qcow

Waldir Pimenta

commit sha c2ca4386515255b5539a4db51ec48ce5708d2a11

LICENSE.md: add title and copyright year range

view details

David Scott

commit sha 76890fd992bb14cee1bfb272885a76e2844414df

Merge pull request #109 from waldyrious/patch-1 LICENSE.md: add title and copyright year range

view details

Etienne Millon

commit sha 7fe7b6506c26587988ebd965c1fc53ad1efc1dc0

Fix warnings in dev mode

view details

Etienne Millon

commit sha befb1f0a0daa7eb5bde20f8474f81bbcec761a94

dune upgrade

view details

Etienne Millon

commit sha 53005a1463c775d01850039692cad461cce6864b

Improve dune files

view details

Etienne Millon

commit sha 23232592549ca71ed9fe1796e3295de8f88e7788

Use dune in Makefile

view details

David Scott

commit sha 7ddec1f8c4336b50bce93579dbae779875962a22

Update to Mirage 4.0 interfaces Also - remove dependency on Nbd for now, as it doesn't support the Mirage 4.0 interfaces - use the stdlib-shims - update to opam2 Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha fb512c19c59821717682fc9d1a36a6e651cba99d

travis: pin mirage-block-ramdisk to master This is needed for the Mirage 4.0 interfaces. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 43f6e0daa3053731ec443e49558651bd38687efa

travis: modernise the .travis.yml This is based on mirage-block-unix's version. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 0fb1e347295714695d8f8edb469e5227d694b130

Merge pull request #112 from emillon/dune Fix warnings in dev mode and port to dune

view details

David Scott

commit sha d914e1924e1787668f3fd5aece8a946bd7e8ac2c

opam: remove unnecessary dep on oasis Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 4b1fdcff9f1b3da26dd830e727da0d6d6bac0daa

opam: remove unnecessary dep on ocamlfind Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 69ccca41dca8cd47d512259b02726a2a8e19cbd0

CHANGES.md: add entries for 0.11.5 Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 0303f78e5a03542ad9fe4a3d6dbaa72cd600abd0

circleci: modernize configuration Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 84478f7f83296271495c193c60cdea2e6af99fd1

Merge pull request #117 from djs55/release.0.11.0 Prepare to release 0.11.0

view details

David Scott

commit sha a3e8038215672b189824c110f11c1b3d14abf289

circleci: remove macos (requires a paid plan) Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 91f49f8b4d134c64aa79b2434c4ff47eaabef09c

Merge pull request #118 from djs55/circleci-remove-macos circleci: remove macos (requires a paid plan)

view details

David Scott

commit sha 0b531ceb992074293fa3b779a5f187d9b58a98d4

circleci: build in a docker image Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 8830059e221daed5cc14ca219babf865517a7dac

Merge pull request #119 from djs55/circleci-use-docker circleci: build in a docker image

view details

David Scott

commit sha c18bc85601980c5a64c021810568831f93605730

circleci: use docker so we can test things locally Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

delete branch djs55/ocaml-qcow

delete branch : circleci-use-docker

delete time in a month

push eventmirage/ocaml-qcow

David Scott

commit sha c18bc85601980c5a64c021810568831f93605730

circleci: use docker so we can test things locally Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 068c8ff6c84d081425513540c102893968b9ac3c

Add dockerignore Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 2b3be4ac02c37216befe01bf667379de4e13eadc

qcow-tool: depend on qcow {= version} Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 21ff6f82bc0613df6bd72307ddc24763521c869c

qcow-tool: remove unnecessary deps on mirage-...-lwt Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 14d8babd2d7c0683bd7976640f052bfff8d1347c

opam: remove jbuilder and ocamlfind Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 9384f9972ee9443f60d1d1e9f4943312182c3543

docker: build both qcow and qcow-tool Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha d98922ae1d0a7f88377b521460ca1759ff46a0bf

docker: set the package versions to "dev" Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha cb3c5afbbfd608c841a16b9a8d9ecde2a82bbb7e

qcow-tool: remove < 2.0.0 on mirage-time I've got mirage-time.2.0.1 installed locally and it seems fine. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha b4766b6e658bc486b510bd249ae25f0a5bbef824

qcow-tool: remove dependency on nbd (not needed any more) Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 4605d655d17d9b61fa1ebe57b46782353962164a

opam: qcow, qcow-tool should have the same mirage-types-lwt constraints Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha f20551b7431e03ce1ed31691a380a542e38f87dd

docker: build manually on alpine:latest This avoids the 1GB opam2 image. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 863b56e2c1cb2031366d5b1ee77af4ff03d39c66

docker: use the correct path for the qcow-tool binary Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 2418c66627dcce8420bcb06d7547db171528060a

Merge pull request #120 from djs55/circleci-use-docker circleci: use docker so we can test things locally

view details

push time in a month

PR merged mirage/ocaml-qcow

circleci: use docker so we can test things locally

Signed-off-by: David Scott dave@recoil.org

+26 -16

5 comments

5 changed files

djs55

pr closed time in a month

pull request commentmirage/ocaml-qcow

circleci: use docker so we can test things locally

Woohoo, we now have a green CI! A fast ocaml-ci and a slow CircleCI. I might disable the slow CircleCI.

djs55

comment created time in a month

push eventdjs55/ocaml-qcow

David Scott

commit sha 863b56e2c1cb2031366d5b1ee77af4ff03d39c66

docker: use the correct path for the qcow-tool binary Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

push eventdjs55/ocaml-qcow

David Scott

commit sha cb3c5afbbfd608c841a16b9a8d9ecde2a82bbb7e

qcow-tool: remove < 2.0.0 on mirage-time I've got mirage-time.2.0.1 installed locally and it seems fine. Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha b4766b6e658bc486b510bd249ae25f0a5bbef824

qcow-tool: remove dependency on nbd (not needed any more) Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 4605d655d17d9b61fa1ebe57b46782353962164a

opam: qcow, qcow-tool should have the same mirage-types-lwt constraints Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha f20551b7431e03ce1ed31691a380a542e38f87dd

docker: build manually on alpine:latest This avoids the 1GB opam2 image. Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

push eventdjs55/ocaml-qcow

David Scott

commit sha d98922ae1d0a7f88377b521460ca1759ff46a0bf

docker: set the package versions to "dev" Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

push eventdjs55/ocaml-qcow

David Scott

commit sha 14d8babd2d7c0683bd7976640f052bfff8d1347c

opam: remove jbuilder and ocamlfind Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 9384f9972ee9443f60d1d1e9f4943312182c3543

docker: build both qcow and qcow-tool Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

push eventdjs55/ocaml-qcow

David Scott

commit sha 2b3be4ac02c37216befe01bf667379de4e13eadc

qcow-tool: depend on qcow {= version} Signed-off-by: David Scott <dave@recoil.org>

view details

David Scott

commit sha 21ff6f82bc0613df6bd72307ddc24763521c869c

qcow-tool: remove unnecessary deps on mirage-...-lwt Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

pull request commentmirage/ocaml-qcow

circleci: use docker so we can test things locally

@talex5 is it possible to run the planning phase of ocaml-ci locally?

djs55

comment created time in a month

pull request commentmirage/ocaml-qcow

circleci: use docker so we can test things locally

@talex5 thanks for taking a look -- I'll fiddle with it a bit and see if I can make it happy!

djs55

comment created time in a month

push eventdjs55/ocaml-qcow

David Scott

commit sha 67205b2f6a1840cbdf2db0aa3122aacddb8552af

qcow-tool: remove version constraint on qcow Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

push eventdjs55/ocaml-qcow

David Scott

commit sha 068c8ff6c84d081425513540c102893968b9ac3c

Add dockerignore Signed-off-by: David Scott <dave@recoil.org>

view details

push time in a month

PR opened mirage/ocaml-qcow

circleci: use docker so we can test things locally

Signed-off-by: David Scott dave@recoil.org

+10 -7

0 comment

2 changed files

pr created time in a month

create barnchdjs55/ocaml-qcow

branch : circleci-use-docker

created branch time in a month

more