-
-
Notifications
You must be signed in to change notification settings - Fork 9.6k
[Messenger] New messenger:stop-workers Command #30754
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
@@ -36,6 +36,10 @@ | |||
<tag name="cache.pool" /> | |||
</service> | |||
|
|||
<service id="cache.messenger" parent="cache.app" public="false"> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm using cache.app
because we need this value to be shared amongst multiple servers and it should persist through deploys. However, there is no precedent that I know if for the core to use cache.app
. And also, perhaps the service id should be more specific to the "messenger restart" - and not just cache.messenger
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can't we use a bus instead here to broadcast to all workers?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If we keep a cache pool, it should be a dedicated one btw, extending from cache.app
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can't we use a bus instead here to broadcast to all workers?
It means we'd have to build a broadcast feature, which we don't have nor plan (yet?) to have.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I've made the cache service id more specific - still extends from cache.app
like before
c0fc67e
to
6ea1002
Compare
Recently |
#30708 has been merged now, so this one can be rebased. |
6ea1002
to
be03f89
Compare
I've kept the longer command name here, because you won't type it often (it's a command to run at deploy mostly) and I think the longer name makes it a bit more descriptive. This is ready for review! It does actually work locally ;) |
be03f89
to
5897162
Compare
Thank you @weaverryan. |
…ryan) This PR was squashed before being merged into the 4.3-dev branch (closes #30754). Discussion ---------- [Messenger] New messenger:stop-workers Command | Q | A | ------------- | --- | Branch? | master | Bug fix? | no | New feature? | yes | BC breaks? | no | Deprecations? | no | Tests pass? | yes | Fixed tickets | Kinda of #29451 | License | MIT | Doc PR | symfony/symfony-docs#11236 o/ me again. This requires and is built on top of #30708 When you deploy, all workers need to be stopped and restarted. That's not currently possible, unless you manually track the pids and send a SIGTERM signal. We can make that much easier :). Now run: ``` bin/console messenger:stop-workers ``` And it will signal to all workers (even if they're distributed on other servers) that they should stop, once they finish processing their current message. This is done via a key in `cache.app`. Cheers! Commits ------- 5897162 [Messenger] New messenger:stop-workers Command
@@ -36,6 +36,10 @@ | |||
<tag name="cache.pool" /> | |||
</service> | |||
|
|||
<service id="cache.messenger.restart_workers_signal" parent="cache.app" public="false"> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
public=false should be removed
$io = new SymfonyStyle($input, $output instanceof ConsoleOutputInterface ? $output->getErrorOutput() : $output); | ||
|
||
$cacheItem = $this->restartSignalCachePool->getItem(StopWhenRestartSignalIsReceived::RESTART_REQUESTED_TIMESTAMP_KEY); | ||
$cacheItem->set(time()); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'd suggest using microtime(true) everywhere
$this->decoratedWorker->stop(); | ||
} | ||
|
||
private function shouldRestart(int $workerStartedAt) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
float for microtime()
This PR was merged into the 4.3-dev branch. Discussion ---------- [Messenger] fix review | Q | A | ------------- | --- | Branch? | master | Bug fix? | yes | New feature? | no | BC breaks? | no | Deprecations? | no | Tests pass? | yes | Fixed tickets | - | License | MIT | Doc PR | - Fixes my review left in #30754 /cc @weaverryan FYI Commits ------- a3de902 [Messenger] fix review
@@ -83,6 +83,9 @@ | |||
<argument type="collection" /> <!-- Receiver names --> | |||
<argument type="service" id="messenger.retry_strategy_locator" /> | |||
<argument type="service" id="event_dispatcher" /> | |||
<call method="setCachePoolForRestartSignal"> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why is this one a call
while it's an argument on the other one? 🤔
…(sroze) This PR was merged into the 4.3-dev branch. Discussion ---------- [Messenger] Add `psr/cache` on Messenger's dependencies | Q | A | ------------- | --- | Branch? | master | Bug fix? | yes | New feature? | no | BC breaks? | no | Deprecations? | no | Tests pass? | yes | Fixed tickets | #30754 | License | MIT | Doc PR | ø Commits ------- 0b5004f Add `psr/cache` on Messenger's dependencies
o/ me again.
This requires and is built on top of #30708
When you deploy, all workers need to be stopped and restarted. That's not currently possible, unless you manually track the pids and send a SIGTERM signal. We can make that much easier :).
Now run:
And it will signal to all workers (even if they're distributed on other servers) that they should stop, once they finish processing their current message. This is done via a key in
cache.app
.Cheers!