TrueCloudLab/frostfs-s3-gw
17
3
Fork 19
Code Issues 28 Pull requests 6 Projects Releases 18 Packages Activity Actions

[#291] server auto re-binding #309

Merged
alexvanin merged 1 commit from pogpp/frostfs-s3-gw:feature/291-rebinding into master 2024-03-27 16:08:29 +00:00
Conversation 1 Commits 1 Files changed 7 +147 -27
pogpp commented 2024-02-11 18:03:11 +00:00
Member
Copy link

Signed-off-by: Pavel Pogodaev p.pogodaev@yadro.com

Signed-off-by: Pavel Pogodaev <p.pogodaev@yadro.com>
pogpp force-pushed feature/291-rebinding from 3a66494805 to 07c5d6a15b 2024-02-12 09:04:15 +00:00 Compare
alexvanin reviewed 2024-02-13 13:15:05 +00:00
alexvanin left a comment
Owner
Copy link

Can we look at one more thing, that is not implemented in storage, as far as I see.

What happens when link is down on one of the connected endpoints?

  1. Can application detect it by looking at the Server structure?
  2. Is it required to reschedule reconnect when link is up again?
  3. Can we do that in runtime?
Can we look at one more thing, that **is not** implemented in storage, as far as I see. What happens when link is down on one of the connected endpoints? 1) Can application detect it by looking at the `Server` structure? 2) Is it required to reschedule reconnect when link is up again? 3) Can we do that in runtime?
cmd/s3-gw/app.go Outdated
@ -983,1 +989,4 @@
}
func (a *App) scheduleReconnect(ctx context.Context) {
a.wg.Add(1)
alexvanin commented 2024-02-13 13:05:44 +00:00
Owner
Copy link

waitgroup is used when application awaits several jobs to be done, e.g.

for job := range jobs {
    wg.Add(1)
    go func() {
        wg.Done()
        // do work
    }
}
wg.Wait() // wait until all jobs are done in go-routines to continue

Storage node awaits all routines in main function.

S3 gateway, on the other hand, does not use wait group, but awaits channel closing.

I suggest to remove waitgroup here and do not await for reconnection in case of closing applications.

waitgroup is used when application awaits several jobs to be done, e.g. ```go for job := range jobs { wg.Add(1) go func() { wg.Done() // do work } } wg.Wait() // wait until all jobs are done in go-routines to continue ``` Storage node [awaits](https://git.frostfs.info/TrueCloudLab/frostfs-node/src/commit/d69d318cb0d3ba8ce2d652606fa1192ac9a42b97/cmd/frostfs-node/main.go#L171) all routines in main function. S3 gateway, on the other hand, does not use wait group, but awaits [channel closing](https://git.frostfs.info/TrueCloudLab/frostfs-s3-gw/src/commit/cc34f659d17c314cc3273023e874b6c73d1744fe/cmd/s3-gw/app.go#L656). I suggest to remove waitgroup here and do not await for reconnection in case of closing applications.
👍 1
alexvanin marked this conversation as resolved
dkirillov reviewed 2024-02-14 08:09:43 +00:00
cmd/s3-gw/app.go Outdated
@ -984,0 +993,4 @@
go func() {
defer a.wg.Done()
t := time.NewTicker(a.cfg.GetDuration(cfgReconnectInterval))
dkirillov commented 2024-02-14 06:59:41 +00:00
Member
Copy link

Why do we use a.cfg? It seems we have already got reconnect interval and save it to settings before.

Now if we don't specify this param in config we get:

panic: non-positive interval for NewTicker

goroutine 296 [running]:
time.NewTicker(0xc001aa5c20?)
        /home/denis/go/go1.20.4/src/time/tick.go:24 +0x10f
main.(*App).scheduleReconnect.func1()
        /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:996 +0x9e
created by main.(*App).scheduleReconnect
        /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:993 +0xaa
Why do we use `a.cfg`? It seems we have already got reconnect interval and save it to settings before. Now if we don't specify this param in config we get: ``` panic: non-positive interval for NewTicker goroutine 296 [running]: time.NewTicker(0xc001aa5c20?) /home/denis/go/go1.20.4/src/time/tick.go:24 +0x10f main.(*App).scheduleReconnect.func1() /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:996 +0x9e created by main.(*App).scheduleReconnect /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:993 +0xaa ```
👍 1
dkirillov commented 2024-02-14 07:53:21 +00:00
Member
Copy link

Let's also use

defer t.Stop()

after ticker initialization

and

t.Reset()

after unsuccessful reconnect

Let's also use ```golang defer t.Stop() ``` after ticker initialization and ```golang t.Reset() ``` after unsuccessful reconnect
👍 1
dkirillov marked this conversation as resolved
cmd/s3-gw/app.go Outdated
@ -984,0 +1016,4 @@
zap.String("tls cert", serverInfo.TLS.CertFile), zap.String("tls key", serverInfo.TLS.KeyFile),
}
srv, err := newServer(ctx, serverInfo)
dkirillov commented 2024-02-14 08:08:03 +00:00
Member
Copy link

It's not enough just create new server listener. We have to serve this listener, see 07c5d6a15b/cmd/s3-gw/app.go (L707-L716)

It's not enough just create new server listener. We have to serve this listener, see https://git.frostfs.info/pogpp/frostfs-s3-gw/src/commit/07c5d6a15be87848d84495e18a83291e5b206003/cmd/s3-gw/app.go#L707-L716
dkirillov commented 2024-02-15 08:12:48 +00:00
Member
Copy link

Still valid

Still valid
dkirillov marked this conversation as resolved
cmd/s3-gw/app.go Outdated
@ -984,0 +1020,4 @@
if err != nil {
a.log.Error(logs.ServerReconnectFailed, zap.Error(err))
return false
dkirillov commented 2024-02-14 07:59:19 +00:00
Member
Copy link

Why we exit function on first error? I think it's better to try other servers reconnect too

Why we exit function on first error? I think it's better to try other servers reconnect too
👍 1
dkirillov marked this conversation as resolved
cmd/s3-gw/app.go Outdated
@ -984,0 +1024,4 @@
}
a.metrics.MarkHealthy(serverInfo.Address)
a.servers = append(a.servers, srv)
dkirillov commented 2024-02-14 08:05:16 +00:00
Member
Copy link

It seems we have to guard a.servers by mutex, because we can read in separate goroutine

It seems we have to guard `a.servers` by mutex, because we [can read in separate goroutine](https://git.frostfs.info/pogpp/frostfs-s3-gw/src/commit/07c5d6a15be87848d84495e18a83291e5b206003/cmd/s3-gw/app.go#L886)
👍 1
dkirillov commented 2024-02-15 08:11:54 +00:00
Member
Copy link

a.serverIndex method still reads a.server without guard. The same for e831c5de82/cmd/s3-gw/app.go (L706)

`a.serverIndex` method still reads `a.server` without guard. The same for https://git.frostfs.info/pogpp/frostfs-s3-gw/src/commit/e831c5de824bd64595afa0a355bb6ffa694d8a38/cmd/s3-gw/app.go#L706
dkirillov marked this conversation as resolved
config/config.yaml Outdated
@ -25,6 +25,8 @@ peers:
priority: 2
weight: 0.9
reconnect_interval: 60000000000
dkirillov commented 2024-02-14 06:50:46 +00:00
Member
Copy link

I believe we can write reconnect_interval: 1m.
Also we should add this param to config.env

I believe we can write `reconnect_interval: 1m`. Also we should add this param to `config.env`
👍 1
dkirillov commented 2024-02-14 08:08:57 +00:00
Member
Copy link

We also have to update docs/configuration.md and changelog

We also have to update `docs/configuration.md` and changelog
👍 1
dkirillov commented 2024-02-15 08:09:51 +00:00
Member
Copy link

config.env still missing

`config.env` still missing
dkirillov commented 2024-02-20 13:27:11 +00:00
Member
Copy link

still valid

still valid
👍 1
dkirillov marked this conversation as resolved
pogpp force-pushed feature/291-rebinding from 07c5d6a15b to 4e5c1905d4 2024-02-14 11:45:05 +00:00 Compare
pogpp force-pushed feature/291-rebinding from 4e5c1905d4 to e831c5de82 2024-02-14 17:40:05 +00:00 Compare
pogpp closed this pull request 2024-02-19 07:23:56 +00:00
pogpp reopened this pull request 2024-02-19 07:24:05 +00:00
pogpp changed title from WIP: [#291] server auto re-binding to [#291] server auto re-binding 2024-02-19 07:24:34 +00:00
dkirillov reviewed 2024-02-20 13:28:08 +00:00
cmd/s3-gw/app.go Outdated
@ -877,6 +885,8 @@ func (a *App) updateServers() error {
}
func (a *App) serverIndex(address string) int {
a.settings.mu.RLock()
dkirillov commented 2024-02-20 13:28:08 +00:00
Member
Copy link

I don't like using mutex from settings. Can we add separate mutext?

I don't like using mutex from settings. Can we add separate mutext?
👍 1
dkirillov marked this conversation as resolved
dkirillov reviewed 2024-02-20 13:29:01 +00:00
cmd/s3-gw/app.go Outdated
@ -969,0 +983,4 @@
}
func (a *App) getServer(index int) Server {
a.settings.mu.RLock()
dkirillov commented 2024-02-20 13:29:01 +00:00
Member
Copy link

Let's accept address here and return nil in case of missing. Then we don't need a.serverIndex method

Let's accept address here and return nil in case of missing. Then we don't need `a.serverIndex` method
👍 1
dkirillov marked this conversation as resolved
dkirillov reviewed 2024-02-20 13:30:04 +00:00
cmd/s3-gw/app.go Outdated
@ -984,0 +1047,4 @@
a.log.Info(logs.StartingServer, zap.String("address", srv.Address()))
if err = sr.Serve(srv.Listener()); err != nil && !errors.Is(err, http.ErrServerClosed) {
a.metrics.MarkUnhealthy(srv.Address())
dkirillov commented 2024-02-20 13:30:04 +00:00
Member
Copy link

Ineffective setting unhealthy because below we invoke a.metrics.MarkHealthy anyway

Ineffective setting unhealthy because below we invoke `a.metrics.MarkHealthy` anyway
👍 1
dkirillov commented 2024-02-20 13:33:32 +00:00
Member
Copy link

Should we invoke srv.Listener().Close() and try again init server on next attempt?

Should we invoke `srv.Listener().Close()` and try again init server on next attempt?
dkirillov marked this conversation as resolved
pogpp force-pushed feature/291-rebinding from 9fd42c1c9e to efc7de7413 2024-02-21 12:35:38 +00:00 Compare
pogpp force-pushed feature/291-rebinding from efc7de7413 to 4b333f5d04 2024-02-21 12:36:30 +00:00 Compare
pogpp force-pushed feature/291-rebinding from 4b333f5d04 to b924d69d83 2024-02-21 12:41:23 +00:00 Compare
dkirillov reviewed 2024-02-26 08:13:08 +00:00
dkirillov left a comment
Member
Copy link

Please, rebase

Please, rebase
cmd/s3-gw/app.go Outdated
@ -707,1 +708,3 @@
a.metrics.MarkUnhealthy(a.servers[i].Address())
for i := range servs {
go func(i int) {
a.log.Info(logs.StartingServer, zap.String("address", servs[i].Address()))
dkirillov commented 2024-02-26 08:12:24 +00:00
Member
Copy link

Actually, in general such using servs[i] isn't safe because in the same time we can update server at i index from other goroutine. But in our current case this cannot happen because in other goroutine we do only append and simple read.

cc @alexvanin

Actually, in general such using `servs[i]` isn't safe because in the same time we can update server at `i` index from other goroutine. But in our current case this cannot happen because in other goroutine we do only append and simple read. cc @alexvanin
alexvanin commented 2024-02-28 11:23:16 +00:00
Owner
Copy link

It is similar to passing pointer of servs[i] structure to goroutine. So I think we are okay with this here.

It is similar to passing pointer of `servs[i]` structure to goroutine. So I think we are okay with this here.
dkirillov marked this conversation as resolved
cmd/s3-gw/app.go Outdated
@ -863,3 +868,2 @@
if serverInfo.TLS.Enabled {
if err := a.servers[index].UpdateCert(serverInfo.TLS.CertFile, serverInfo.TLS.KeyFile); err != nil {
if err := a.getServer(serverInfo.Address).UpdateCert(serverInfo.TLS.CertFile, serverInfo.TLS.KeyFile); err != nil {
dkirillov commented 2024-02-26 08:01:15 +00:00
Member
Copy link

Here we must check a.getServer for nil otherwise we get panic:

2024-02-26T10:59:15.575+0300    info    s3-gw/app.go:781        SIGHUP config reload started
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x28 pc=0xec2ad9]

goroutine 281 [running]:
main.(*App).updateServers(0xc000021900)
        /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:869 +0x99
main.(*App).configReload(0xc000021900, {0x156bc58, 0xc00012efc0})
        /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:796 +0x365
main.(*App).Serve(0xc000021900, {0x156bc58, 0xc00012efc0})
        /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:740 +0x68f
created by main.main in goroutine 1
        /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/main.go:17 +0x105
Here we must check `a.getServer` for `nil` otherwise we get `panic`: ``` 2024-02-26T10:59:15.575+0300 info s3-gw/app.go:781 SIGHUP config reload started panic: runtime error: invalid memory address or nil pointer dereference [signal SIGSEGV: segmentation violation code=0x1 addr=0x28 pc=0xec2ad9] goroutine 281 [running]: main.(*App).updateServers(0xc000021900) /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:869 +0x99 main.(*App).configReload(0xc000021900, {0x156bc58, 0xc00012efc0}) /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:796 +0x365 main.(*App).Serve(0xc000021900, {0x156bc58, 0xc00012efc0}) /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/app.go:740 +0x68f created by main.main in goroutine 1 /home/denis/github/tcl/neofs-s3-gw/cmd/s3-gw/main.go:17 +0x105 ```
👍 1
dkirillov marked this conversation as resolved
cmd/s3-gw/app.go Outdated
@ -877,6 +881,8 @@ func (a *App) updateServers() error {
}
func (a *App) serverIndex(address string) int {
dkirillov commented 2024-02-26 08:00:18 +00:00
Member
Copy link

This method isn't being used anymore. Can be dropped

This method isn't being used anymore. Can be dropped
👍 1
dkirillov marked this conversation as resolved
pogpp force-pushed feature/291-rebinding from b924d69d83 to 8a87d9b15b 2024-02-27 10:46:08 +00:00 Compare
pogpp force-pushed feature/291-rebinding from 8a87d9b15b to 777ae514bc 2024-02-27 10:46:58 +00:00 Compare
dkirillov approved these changes 2024-02-27 12:24:00 +00:00
dkirillov left a comment
Member
Copy link

LGTM, but we should decide something about #309 (comment)

LGTM, but we should decide something about https://git.frostfs.info/TrueCloudLab/frostfs-s3-gw/pulls/309#issuecomment-34098
alexvanin commented 2024-02-28 11:24:15 +00:00
Owner
Copy link

I am gonna merge it after v0.29.0-rc.1 tag, because I want to have this feature in the same HTTP and S3 releases. HTTP is already passed across v0.29.0-rc.1.

I am gonna merge it after `v0.29.0-rc.1` tag, because I want to have this feature in the same HTTP and S3 releases. HTTP is already passed across `v0.29.0-rc.1`.
dkirillov requested changes 2024-02-29 09:08:48 +00:00
dkirillov left a comment
Member
Copy link

The scenario I want to pass:

  1. Start s3-gw with the following config:
server:
  - address: 0.0.0.0:8084
    tls:
      enabled: true
      cert_file: /valid/path/to/cert.pem
      key_file: /valid/path/to/key.pem
  - address: 0.0.0.0:8185
    tls:
      enabled: true
      cert_file: /not/existing/path/to/cert.pem
      key_file: /not/existing/path/to/key.pem
  1. Change config and send SIGHUP
server:
  - address: 0.0.0.0:8084
    tls:
      enabled: true
      cert_file: /valid/path/to/cert.pem
      key_file: /valid/path/to/key.pem
  - address: 0.0.0.0:8185
    tls:
      enabled: true
      cert_file: /valid/path/to/cert.pem
      key_file: /valid/path/to/key.pem
  1. Reconnect must be succesfull
The scenario I want to pass: 1. Start s3-gw with the following config: ```yaml server: - address: 0.0.0.0:8084 tls: enabled: true cert_file: /valid/path/to/cert.pem key_file: /valid/path/to/key.pem - address: 0.0.0.0:8185 tls: enabled: true cert_file: /not/existing/path/to/cert.pem key_file: /not/existing/path/to/key.pem ``` 2. Change config and send `SIGHUP` ```yaml server: - address: 0.0.0.0:8084 tls: enabled: true cert_file: /valid/path/to/cert.pem key_file: /valid/path/to/key.pem - address: 0.0.0.0:8185 tls: enabled: true cert_file: /valid/path/to/cert.pem key_file: /valid/path/to/key.pem ``` 3. Reconnect must be succesfull
cmd/s3-gw/app.go Outdated
@ -73,2 +73,3 @@
servers []Server
servers []Server
unbindServers []ServerInfo
dkirillov commented 2024-02-29 08:54:14 +00:00
Member
Copy link

It seems we have to update unbindServers on SIGHUP, similar to a.updateServer.

cc @alexvanin

It seems we have to update `unbindServers` on `SIGHUP`, similar to `a.updateServer`. cc @alexvanin
👍 1
dkirillov marked this conversation as resolved
cmd/s3-gw/app.go Outdated
@ -984,0 +1031,4 @@
zap.String("tls cert", serverInfo.TLS.CertFile), zap.String("tls key", serverInfo.TLS.KeyFile),
}
srv, err := newServer(ctx, serverInfo)
dkirillov commented 2024-02-29 09:06:16 +00:00
Member
Copy link

We will get error here in the following scenario.
Let's say we have the following config:

server:
  - address: 0.0.0.0:8084
    tls:
      enabled: true
      cert_file: /valid/path/to/cert.pem
      key_file: /valid/path/to/key.pem
  - address: 0.0.0.0:8185
    tls:
      enabled: true
      cert_file: /not/existing/path/to/cert.pem
      key_file: /not/existing/path/to/key.pem

Then we successfully start one server and will try reconnect to another:

2024-02-29T11:57:45.180+0300    info    s3-gw/app.go:855        add server      {"address": "0.0.0.0:8084", "tls enabled": true, "tls cert": "/valid/path/to/cert.pem", "tls key": "/valid/path/to/key.pem"}
2024-02-29T11:57:45.180+0300    warn    s3-gw/app.go:849        failed to add server    {"address": "0.0.0.0:8185", "tls enabled": true, "tls cert": "/not/existing/path/to/cert.pem", "tls key": "/not/existing/path/to/key.pem", "error": "failed to update cert: cannot load TLS key pair from certFile '/not/existing/path/to/cert.pem' and keyFile '/not/existing/path/to/key.pem': open /not/existing/path/to/cert.pem: no such file or directory"}

but on reconnect we get:

2024-02-29T11:58:00.181+0300    info    s3-gw/app.go:1025       reconnecting server...
2024-02-29T11:58:00.181+0300    warn    s3-gw/app.go:1036       failed to reconnect server      {"error": "could not prepare listener: listen tcp 0.0.0.0:8185: bind: address already in use"}

This is because we don't close listerer on error here.

After some trying listener be closed (by timeout I suppose, or by GC):

2024-02-29T12:01:45.193+0300    warn    s3-gw/app.go:1036       failed to reconnect server      {"error": "failed to update cert: cannot load TLS key pair from certFile '/not/existing/path/to/cert.pem' and keyFile '/not/existing/path/to/key.pem': open /not/existing/path/to/cert.pem: no such file or directory"}

but it would be nice explicitly close the listener. (Thank Svace)

We will get error here in the following scenario. Let's say we have the following config: ```yaml server: - address: 0.0.0.0:8084 tls: enabled: true cert_file: /valid/path/to/cert.pem key_file: /valid/path/to/key.pem - address: 0.0.0.0:8185 tls: enabled: true cert_file: /not/existing/path/to/cert.pem key_file: /not/existing/path/to/key.pem ``` Then we successfully start one server and will try reconnect to another: ``` 2024-02-29T11:57:45.180+0300 info s3-gw/app.go:855 add server {"address": "0.0.0.0:8084", "tls enabled": true, "tls cert": "/valid/path/to/cert.pem", "tls key": "/valid/path/to/key.pem"} 2024-02-29T11:57:45.180+0300 warn s3-gw/app.go:849 failed to add server {"address": "0.0.0.0:8185", "tls enabled": true, "tls cert": "/not/existing/path/to/cert.pem", "tls key": "/not/existing/path/to/key.pem", "error": "failed to update cert: cannot load TLS key pair from certFile '/not/existing/path/to/cert.pem' and keyFile '/not/existing/path/to/key.pem': open /not/existing/path/to/cert.pem: no such file or directory"} ``` but on reconnect we get: ``` 2024-02-29T11:58:00.181+0300 info s3-gw/app.go:1025 reconnecting server... 2024-02-29T11:58:00.181+0300 warn s3-gw/app.go:1036 failed to reconnect server {"error": "could not prepare listener: listen tcp 0.0.0.0:8185: bind: address already in use"} ``` This is because we don't close [listerer](https://git.frostfs.info/pogpp/frostfs-s3-gw/src/commit/777ae514bcfc4af035de4cfebcfa413e4a2bc14c/cmd/s3-gw/server.go#L60) on error [here](https://git.frostfs.info/pogpp/frostfs-s3-gw/src/commit/777ae514bcfc4af035de4cfebcfa413e4a2bc14c/cmd/s3-gw/server.go#L71). After some trying listener be closed (by timeout I suppose, or by GC): ``` 2024-02-29T12:01:45.193+0300 warn s3-gw/app.go:1036 failed to reconnect server {"error": "failed to update cert: cannot load TLS key pair from certFile '/not/existing/path/to/cert.pem' and keyFile '/not/existing/path/to/key.pem': open /not/existing/path/to/cert.pem: no such file or directory"} ``` but it would be nice explicitly close the listener. (Thank Svace)
dkirillov marked this conversation as resolved
pogpp force-pushed feature/291-rebinding from 777ae514bc to d3e1753438 2024-03-25 12:00:22 +00:00 Compare
pogpp force-pushed feature/291-rebinding from d3e1753438 to e1ae3f3c29 2024-03-26 07:52:05 +00:00 Compare
dkirillov reviewed 2024-03-26 13:02:48 +00:00
cmd/s3-gw/app.go Outdated
@ -977,0 +1045,4 @@
a.log.Info(logs.ServerReconnecting)
var failedServers []ServerInfo
for _, serverInfo := range a.unbindServers {
dkirillov commented 2024-03-26 13:02:48 +00:00
Member
Copy link

Since we update a.unbindServers on SIGHUP we must protect them here by mutex

Since we update `a.unbindServers` on SIGHUP we must protect them here by mutex
dkirillov marked this conversation as resolved
pogpp force-pushed feature/291-rebinding from e1ae3f3c29 to b1ad21edc7 2024-03-26 15:27:16 +00:00 Compare
dkirillov reviewed 2024-03-27 06:38:39 +00:00
cmd/s3-gw/app.go Outdated
@ -977,0 +1066,4 @@
a.log.Info(logs.ServerReconnectedSuccessfully, fields...)
}
a.unbindServers = failedServers
dkirillov commented 2024-03-27 06:38:39 +00:00
Member
Copy link

We must protect this too

We must protect this too
👍 1
dkirillov marked this conversation as resolved
dkirillov reviewed 2024-03-27 06:39:41 +00:00
cmd/s3-gw/app.go Outdated
@ -977,0 +1040,4 @@
a.log.Info(logs.ServerReconnecting)
var failedServers []ServerInfo
servers := a.getUnbindServers()
dkirillov commented 2024-03-27 06:39:41 +00:00
Member
Copy link

We cannot just get all servers by mutex and then read them because in updateUnbindServerInfo we update single i-th item (not whole slice)

We cannot just get all servers by mutex and then read them because in `updateUnbindServerInfo` we update single i-th item (not whole slice)
👍 1
dkirillov marked this conversation as resolved
pogpp force-pushed feature/291-rebinding from b1ad21edc7 to 937407a6da 2024-03-27 08:28:13 +00:00 Compare
pogpp force-pushed feature/291-rebinding from 937407a6da to 7d9ba1a87f 2024-03-27 08:44:14 +00:00 Compare
pogpp force-pushed feature/291-rebinding from 7d9ba1a87f to 573fe68fe2 2024-03-27 09:22:26 +00:00 Compare
pogpp force-pushed feature/291-rebinding from 573fe68fe2 to 32cd3d0b92 2024-03-27 10:34:10 +00:00 Compare
dkirillov approved these changes 2024-03-27 11:18:05 +00:00
pogpp force-pushed feature/291-rebinding from 32cd3d0b92 to bfcde09f07 2024-03-27 11:29:03 +00:00 Compare
dkirillov approved these changes 2024-03-27 14:27:34 +00:00
dkirillov requested review from storage-services-developers 2024-03-27 14:27:40 +00:00
dkirillov requested review from storage-services-committers 2024-03-27 14:27:41 +00:00
alexvanin approved these changes 2024-03-27 16:08:22 +00:00
alexvanin merged commit bfcde09f07 into master 2024-03-27 16:08:29 +00:00
alexvanin deleted branch feature/291-rebinding 2024-03-27 16:08:29 +00:00
Sign in to join this conversation.
Reviewers
No reviewers
dkirillov
alexvanin
TrueCloudLab/storage-services-developers
Labels
Clear labels   
P0

Highest

  
P1

High

  
P2

Medium

  
P3

Low

  
good first issue

Good for newcomers

  
Infrastructure

Infrastructure, Automation, CI/CD or DevOps related stuff

  
blocked

The issue or PR is blocked by something

  
bug

Something is not working

  
config

Configuration format update or breaking change

  
discussion

Talking about something in order to reach a decision or to exchange ideas

  
documentation

Documentation related

  
duplicate

This issue or pull request already exists

  
enhancement

New feature

  
go

Language features adoption

  
help wanted

Extra attention is needed

  
internal

   
invalid

Something is wrong

  
kludge

This is a kludge and we know it

  
observability

Related to metrics, tracing and logs

  
perfomance

Perfomance related

  
question

More information is needed

  
refactoring

Refactoring

  
wontfix

This won't be fixed

No labels
P0
P1
P2
P3
good first issue
Infrastructure
blocked
bug
config
discussion
documentation
duplicate
enhancement
go
help wanted
internal
invalid
kludge
observability
perfomance
question
refactoring
wontfix
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
3 participants
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference: TrueCloudLab/frostfs-s3-gw#309
No description provided.
Delete branch "pogpp/frostfs-s3-gw:feature/291-rebinding"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?