push eventerikh/sdd
commit sha 22706cd25586f67e3173fa3d1f5366e3cc7120f4
Initial spike on heartbeating support Signed-off-by: Erik Hollensbe <[email protected]>
push time in 3 hours
push eventerikh/sdd
commit sha 655aa523c3b18e2d9e81d5ef30bb7423cc35b8cd
Testing for TTLs Signed-off-by: Erik Hollensbe <[email protected]>
push time in a day
push eventerikh/sdd
commit sha 8bd93cf0516dea66544ce885e54a37a044d09a11
TTL support for mDNS; no tests yet. Signed-off-by: Erik Hollensbe <[email protected]>
push time in a day
startedmicro/go-micro
started time in a day
push eventerikh/sdd
commit sha 9de3501acd95c63f46aad04291526ea0444076da
Corrected resolver monkeying in tests, after changes to domain usage (the API changed) Signed-off-by: Erik Hollensbe <[email protected]>
push time in a day
push eventerikh/sdd
commit sha 7151c011911d0ab09b0df507669c88694b034b02
Updates to client workings with domains: The default domain is now `local`, not `sdd`. This is to comply with multicast DNS semantics by default, which is our intent to do throughout the project. When queries are issued; the domain is automatically appended in the proper format. client.NewClient now requires this domain name be passed; if it is not supplied, the default domain of `local` will be used. multicast discovery is not working right now when local is not in use, I need to read more of the spec to fully understand that problem. Signed-off-by: Erik Hollensbe <[email protected]>
push time in a day
push eventerikh/dnsserver
commit sha a686a756956691cda05c39d41492470785b076a7
Update README to remove stupid Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha 071e192dfba1a673ea96b68a51e8eb0787e54441
impelmentation of GRPCDial Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha 7962e23afa04a6800263049a1e08eb144e51e00b
Tweaks for client Dial method; no tests yet. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha af58c35fa465911556ef43e4c84d9c86a9912974
additional test coverage Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha cd2eb5ee1520cf02ff1d36d2bde3fb3f00c9fd08
Many upgrades to the client around discovery of port-tagged services via SRV records Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha 56f8b039f5a3b7f8fc9631bbd350fbeff4579138
Unregister method; tests Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha f254fed56dbbfe3bfb902e60b45659b14142adda
More client test improvements, no new tests yet Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha 18d03562457556cd043f37dea8a3ce348e1220db
Some rework around contexts to make further client tests easier to write Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha bdd92130c352f81d797e45e939024abf86057a6f
signals Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha f43982d8d10c5b3e9db672d7601678e34556b1a2
Upgrade DNS server, utilize new close method Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha 087582d49b2a947de84b0c48dd97aa5bbcc847a8
Fix context use in a few situations Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/dnsserver
commit sha 163211cec01096d5ca8cbb8ad0bffc41782f209e
Remove example, update README Signed-off-by: Erik Hollensbe <[email protected]>
commit sha 503cbdfe7c27cb2e22ac968504fb5ce2f0f703bf
Update LICENSE to reflect fork Signed-off-by: Erik Hollensbe <[email protected]>
commit sha cebb54d60b9ab17e587206cbabaf7cd35f302a04
Support for shutting down the server directly Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha c7e27fc819235c9be5566d1f970b86a390d28cec
Crude, basic context support for server. dnsserver patches need to be made for it to be correct. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha 020831f5d72252e9b29946be6dd031edac235730
tinyci support Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha ddd66f9947774f4d8895116e2002ff3a37b9797a
Some initial testing Signed-off-by: Erik Hollensbe <[email protected]>
push time in 2 days
push eventerikh/sdd
commit sha a7ce80c478c41dec360c3a058eb595e2c65ca95a
running in containers Signed-off-by: Erik Hollensbe <[email protected]>
push time in 3 days
push eventerikh/sdd
commit sha fad0602dcf25efa8f26a5e9d1b0019aecbe9eba0
Build a box build plan Signed-off-by: Erik Hollensbe <[email protected]>
push time in 3 days
push eventerikh/migrator
commit sha 9485ee10af48c43021006bcaaa82b3ba56315ec1
Update golang in task.yml Signed-off-by: Erik Hollensbe <[email protected]>
push time in 4 days
push eventerikh/dnsserver
commit sha 17f8a9fbdf4f29f635416225290c7d7d21d23172
tinyci bootstrap Signed-off-by: Erik Hollensbe <[email protected]>
push time in 4 days
push eventerikh/dnsserver
commit sha 68ba2ce0369517d167360f1362c2a202245fdf60
Some fixups Signed-off-by: Erik Hollensbe <[email protected]>
push time in 5 days
PR opened micro/go-micro
Only shows in travis
Signed-off-by: Erik Hollensbe [email protected]
pr created time in 5 days
pull request commentmicro/go-micro
Signal handling try #2: through the context
PTAL
comment created time in 5 days
push eventerikh/go-micro
commit sha 096e4048cae45cedf8fba04de6dd33e6434cdd43
Now specify HandleSignal as an option to toggle signal handling. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 5 days
push eventerikh/go-micro
commit sha 134ad648cca8cabb96824f6cd27c8df3a5ab6e8e
Write a test to prove memory registry TTLs are busted Signed-off-by: Erik Hollensbe <[email protected]>
commit sha 28c3407424d98dde988f53b6c075e7ba5284d5c2
Additional memory testing and fixups: * DefaultTTL removed * When TTL == 0, it is automatically removed from expiry conditions * Additional improvements to new tests Signed-off-by: Erik Hollensbe <[email protected]>
push time in 5 days
PR opened micro/go-micro
This is #801 (from @milosgajdos83) with the following additions:
- DefaultTTL is not a thing anymore
- Additional tests from #942
- TTL == 0 == never expires
pr created time in 5 days
Pull request review commentmicro/go-micro
Signal handling try #2: through the context
func (s *service) Run() error { return err } - ch := make(chan os.Signal, 1)- signal.Notify(ch, syscall.SIGTERM, syscall.SIGINT, syscall.SIGQUIT)-- select {- // wait on kill signal- case <-ch:- // wait on context cancel- case <-s.opts.Context.Done():+ if s.opts.DisableSignalHandler {+ <-s.opts.Context.Done()
Your comments: https://github.com/micro/go-micro/pull/943#issuecomment-554111145 am I confused?
comment created time in 5 days
Pull request review commentmicro/go-micro
Signal handling try #2: through the context
func (s *service) Run() error { return err } - ch := make(chan os.Signal, 1)- signal.Notify(ch, syscall.SIGTERM, syscall.SIGINT, syscall.SIGQUIT)-- select {- // wait on kill signal- case <-ch:- // wait on context cancel- case <-s.opts.Context.Done():+ if s.opts.DisableSignalHandler {+ <-s.opts.Context.Done()
this is literally what I was doing in #943. Which one do you prefer?
comment created time in 5 days
Pull request review commentmicro/go-micro
Signal handling try #2: through the context
func Context(ctx context.Context) Option { } } +// SignalHandler toggles automatic installation of the signal handler.+// Users of this feature to disable the signal handler, should control liveness+// of the service through the context.+func SignalHandler(b bool) Option {
sg
comment created time in 5 days
push eventerikh/go-micro
commit sha 6b990591f72a119ce8a4c9022dd967ca92a616d6
Write a test to prove memory registry TTLs are busted Signed-off-by: Erik Hollensbe <[email protected]>
commit sha 495190350aadda556623abdf16e69810323eae5f
Merge branch 'memory-registry-ttl' into memory-registry
commit sha 688442713a7fa7af3189754273363c84ebf75042
Additional memory testing and fixups: * DefaultTTL removed * When TTL == 0, it is automatically removed from expiry conditions * Additional improvements to new tests Signed-off-by: Erik Hollensbe <[email protected]>
push time in 5 days
pull request commentmicro/go-micro
Signal handling try #2: through the context
I called it signalhandler -- I felt signal was confusing. I hope that's ok. Otherwise I think it's what you want.
comment created time in 5 days
push eventerikh/go-micro
commit sha 70633f1fb352f2c249b52a444777720725e21a80
Signal handling try #2: through the context This approach expects the user to control their context to manage the service. A signal handler is installed by default by you can specify the option `SignalHandler()` to turn it off. The choice of keeping the signal handler in was intended to preserve backwards compatibility so that users of go-micro today were not surprised by the signal handling changes. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 5 days
pull request commentmicro/go-micro
Signal handling try #2: through the context
nvm, I think I understand. I'll get to it tonight
comment created time in 6 days
pull request commentmicro/go-micro
Signal handling try #2: through the context
So we don't care about backwards compat at all?
comment created time in 6 days
pull request commentmicro/go-micro
Signal handling try #2: through the context
cc @asim
comment created time in 6 days
PR opened micro/go-micro
This approach expects the user to control their context to manage the
service. A signal handler is installed by default by you can specify the
option NoSignalHandler() to turn it off.
The choice of keeping the signal handler in was intended to preserve backwards compatibility so that users of go-micro today were not surprised by the signal handling changes.
Signed-off-by: Erik Hollensbe [email protected]
pr created time in 6 days
pull request commentmicro/go-micro
SignalChan provides a way to override micro's signal behavior
this is exactly what I suggested in slack before. Would you like me to do that patch instead?
On Thu, Nov 14, 2019 at 2:30 PM Asim Aslam [email protected] wrote:
No we can't really merge this right now. I'm not sure we should leak something like a signal channel out. It might be more appropriate to just operate on Context in the event it's passed in and ignore use of signal in that specific case. Otherwise we're always going to be tied to some option that allows someone to pass in a signal channel for interrupt. I just have a bad feeling it's going to be abused.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/micro/go-micro/pull/943?email_source=notifications&email_token=AAAET2YZIE2GNXMRXV7XETDQTXGOZA5CNFSM4JNERT5KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEDRBKI#issuecomment-554111145, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAET23U5C5SNSPN4U2H4VDQTXGOZANCNFSM4JNERT5A .
comment created time in 7 days
pull request commentmicro/go-micro
SignalChan provides a way to override micro's signal behavior
that should be fixed now
comment created time in 7 days
push eventerikh/go-micro
commit sha 61206e8f0f636705bab76d6ccdd763c3ef524910
SignalChan provides a way to override micro's signal behavior If provided, it will use this provided channel and not install a signal handler of its own Signed-off-by: Erik Hollensbe <[email protected]>
push time in 7 days
push eventerikh/go-micro
commit sha 18ca90a54b631a7bfa28779099911eff0ce8d182
SignalChan provides a way to override micro's signal behavior If provided, it will use this provided channel and not install a signal handler of its own. This is the simplest approach I could think of. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 7 days
pull request commentmicro/go-micro
SignalChan provides a way to override micro's signal behavior
https://github.com/micro/go-micro/pull/943/files#diff-4aae454123f50b906f9aaef44f10011eR169-R173 for more information.
comment created time in 7 days
pull request commentmicro/go-micro
SignalChan provides a way to override micro's signal behavior
No notifications are registered on the channel; it's basically a no-op if you pass nil. Then you can handle signals however you'd like. If you want to register a specific set of signals to the channel, then call signal.Notify yourself on a channel you pass.
There is a small bug in this, the default value of SignalChan on the options struct is not nil, which means the signal handler will never get installed properly right now. I need to fix that, so don't merge this until after that.
comment created time in 7 days
pull request commentmicro/go-micro
SignalChan provides a way to override micro's signal behavior
I documented the flag, fwiw.
On Thu, Nov 14, 2019 at 10:18 AM Erik Hollensbe [email protected] wrote:
pass nil to the option
On Thu, Nov 14, 2019 at 4:49 AM Vasiliy Tolstov [email protected] wrote:
@asim https://github.com/asim i'm prefer to have ability to disable signal processing inside server at all. what you think? @erikh https://github.com/erikh how can i disable singla processing with such option? I need to create new non nil signal channel and pass to service?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/micro/go-micro/pull/943?email_source=notifications&email_token=AAAET23FYXNIVWW5CITCXETQTVCNHA5CNFSM4JNERT5KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEBXA3A#issuecomment-553873516, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAET25B6OBOJXPTYIZM4LTQTVCNHANCNFSM4JNERT5A .
comment created time in 7 days
pull request commentmicro/go-micro
SignalChan provides a way to override micro's signal behavior
pass nil to the option
On Thu, Nov 14, 2019 at 4:49 AM Vasiliy Tolstov [email protected] wrote:
@asim https://github.com/asim i'm prefer to have ability to disable signal processing inside server at all. what you think? @erikh https://github.com/erikh how can i disable singla processing with such option? I need to create new non nil signal channel and pass to service?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/micro/go-micro/pull/943?email_source=notifications&email_token=AAAET23FYXNIVWW5CITCXETQTVCNHA5CNFSM4JNERT5KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEBXA3A#issuecomment-553873516, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAET25B6OBOJXPTYIZM4LTQTVCNHANCNFSM4JNERT5A .
comment created time in 7 days
startedcontainerd/release-tool
started time in 7 days
push eventerikh/go-micro
commit sha 3e30b088368409e7e0e5bbc98cdf5d3c39233e58
SignalChan provides a way to override micro's signal behavior If provided, it will use this provided channel and not install a signal handler of its own. This is the simplest approach I could think of. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 7 days
push eventerikh/go-micro
commit sha 9a7bb8e2449bbff05a2e6b831c207406d3c55ec5
SignalChan provides a way to override micro's signal behavior If provided, it will use this provided channel and not install a signal handler of its own. This is the simplest approach I could think of. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 7 days
PR opened micro/go-micro
If provided, it will use this provided channel and not install a signal handler of its own.
This is the simplest approach I could think of.
Signed-off-by: Erik Hollensbe [email protected]
pr created time in 7 days
pull request commentmicro/go-micro
Write a test to prove memory registry TTLs are busted
I've included an alternative implementation called the "ContextRegistry". This registry uses golang contexts and context.WithTimeout to implement TTLs; these TTLs are then resolved at querying time instead of the background. Seems to be efficient enough and I doubt it will break due to a timing issue.
comment created time in 8 days
push eventerikh/go-micro
commit sha 870ca7ceb54d5ec080602c31e8c884908d794dda
Memory registry with context-backed TTLs This registry uses the golang context package to provide TTLs, otherwise the implementation is largely the same. Included are some additional concurrency tests. Signed-off-by: Erik Hollensbe <[email protected]>
push time in 8 days
pull request commentmicro/go-micro
Write a test to prove memory registry TTLs are busted
Discussions in slack lead me to think that we can resolve this with contexts that work like promises more or less; contexts are registered with context.WithTimeout() and could be simply iterated over for a given service and Done() could be checked to see if the record is expired or not, at which point it can be pruned at query time if done safely.
WDYT?
comment created time in 8 days
pull request commentmicro/go-micro
Write a test to prove memory registry TTLs are busted
I've updated the description of the ticket with more details; sorry, was in a rush.
comment created time in 8 days
pull request commentmicro/go-micro
Write a test to prove memory registry TTLs are busted
(which it doesn't appear to be doing)
comment created time in 8 days
pull request commentmicro/go-micro
Write a test to prove memory registry TTLs are busted
ignore the second commit, please. it's largely to demonstrate the code does work if the timings are correct.
comment created time in 8 days
push eventerikh/go-micro
commit sha bd34bfa758c5036496dc36838ae84dd5d43af813
Write a test to prove memory registry TTLs are busted Signed-off-by: Erik Hollensbe <[email protected]>
commit sha d2b35a575ab2593ef396c337a679d8deeb6386e1
fixes to make it pass -- DO NOT MERGE Signed-off-by: Erik Hollensbe <[email protected]>
push time in 8 days
pull request commentmicro/go-micro
Write a test to prove memory registry TTLs are busted
cc @milosgajdos83
comment created time in 8 days
PR opened micro/go-micro
Signed-off-by: Erik Hollensbe [email protected]
pr created time in 8 days
push eventerikh/go-micro
commit sha 79da1e11c97b08d3c8162716149500c40490fb79
Write a test to prove memory registry TTLs are busted Signed-off-by: Erik Hollensbe <[email protected]>
push time in 8 days
issue openedmicro/protoc-gen-micro
Use of protobuf extended types causes unnecessary import
Here's an approximation of my service:
syntax = "proto3";
package assetsvc;
import "google/protobuf/empty.proto";
service Assetsvc {
rpc PutLog (stream LogSend) returns (google.protobuf.Empty); // PutLog sends a log
}
message LogSend {
}
The result after a build with this protobuf is this:
service/assetsvc/proto/assetsvc/assetsvc.micro.go:9:2: imported and not used: "github.com/golang/protobuf/ptypes/empty"
Thanks for making an interesting tool!
created time in 10 days
push eventerikh/migrator
commit sha 70396226fc0eee85a9f5310c92f3640fd51382c6
Upgrade tooling / Fix recovery issue / Fix testing env This repository's been suffering a bit of bit-rot, and when I found a bug in the migrations system I needed to fix, I took some time to do a few other things. The core issue is that if you have a series of migrations 0-3, and 3 is bad and aborts the migration, the current code would have you start back at 0 the next run, which is wrong since all the way up to 2 have completed. An ugly oversight that I hope never returns. Tests were added for this bug. Other things: upgrade all go tooling, upgrade to latest (future proofed) postgres packaging. Signed-off-by: Erik Hollensbe <[email protected]>
commit sha 1b1bfa30085493aee5bf4715921dc24d1d9e6837
Merge branch 'polish'
push time in 11 days
push eventtinyci/ci-agents
commit sha 6a015ea8349aa2d835b68419e9f66c2367522ca7
Fix issues with branch creation leading to ugly refs on the UI Signed-off-by: Erik Hollensbe <[email protected]>
commit sha 462c943e949bda38321041a8817592f8769b3892
Merge pull request #173 from erikh/fix-branches Fix issues with branch creation leading to ugly refs on the UI
push time in 12 days
PR merged tinyci/ci-agents
Signed-off-by: Erik Hollensbe [email protected]
pr closed time in 12 days
PR opened tinyci/ci-agents
Signed-off-by: Erik Hollensbe [email protected]
pr created time in 12 days
push eventtinyci/ci-demo
commit sha b3ee9f324c67b997d2d6707e25876a42d8ca3e61
Update to 0.2.1 Signed-off-by: Erik Hollensbe <[email protected]>
push time in 15 days
push eventtinyci/ci-deploy
commit sha 1ffc55c90220a1f38a78729519e6f4c88c630c9a
Update to 0.2.1 Signed-off-by: Erik Hollensbe <[email protected]>
push time in 15 days
push eventtinyci/tinyci
commit sha 124bc54f680f69f1c5c9f9f088d2f4ef2d6525da
Oops Signed-off-by: Erik Hollensbe <[email protected]>
push time in 15 days
push eventtinyci/ci-agents
commit sha 6f87c6d04b978d6fc8713f97234571cdf8a547e8
Bump version for 0.2.1 release Signed-off-by: Erik Hollensbe <[email protected]>
push time in 15 days
push eventtinyci/tinyci
commit sha 4fea9263541ad739e966c65bcfd694c585b933da
Release 0.2.1 Signed-off-by: Erik Hollensbe <[email protected]>
push time in 15 days
push eventtinyci/ci-runners
commit sha 2bbab625a26b3daa079b3e56a8471696ecb68926
tinyCI 0.2.1 Signed-off-by: Erik Hollensbe <[email protected]>
push time in 15 days
push eventtinyci/ci-runners
commit sha 66280a6d05e3f2d2cd8b4bbf0dfb9d98509b9f91
Fix a goroutine leak in the runner signal handlers Signed-off-by: Erik Hollensbe <[email protected]>
commit sha 2905afb06904e2653b9a7bcee03214d02d4c1476
Merge pull request #50 from erikh/fix-goroutine-leak Fix a goroutine leak in the runner signal handlers
push time in 15 days
PR merged tinyci/ci-runners
Signed-off-by: Erik Hollensbe [email protected]
pr closed time in 15 days
PR opened tinyci/ci-runners
Signed-off-by: Erik Hollensbe [email protected]
pr created time in 15 days
push eventtinyci/ci-demo
commit sha 5bb28be8245f3eaa9d1310dd9a65ee4f1d69a762
Upgrade demo stack to include 0.2.0 Signed-off-by: Erik Hollensbe <[email protected]>
push time in 16 days
push eventtinyci/ci-deploy
commit sha c8b7f98c6041d1da748c31acc2097d2f75e1991a
update apt cache before doing anything with it Signed-off-by: Erik Hollensbe <[email protected]>
push time in 16 days