Cls issue 70 #101
Cls issue 70 #101
Conversation
|
@crossjam thanks for taking a stab at this! |
| except Exception, e: | ||
| logger.warning('[%s:%s] error sending CLS, closing', | ||
| conn.id, self.name) | ||
| conn.close() |
There was a problem hiding this comment.
we need to add accounting for outstanding messages and wait until they're all responded to before we actually close the connection (after sending CLS)
There was a problem hiding this comment.
Is this outstanding message accounting an issue for just Writers? Readers? Or both?
Also, is an outstanding count good enough, or do individual messages need to be tracked?
There was a problem hiding this comment.
just the reader - a count is sufficient.
There was a problem hiding this comment.
How about I boost this up to the conn level, where there's already a counter for in-flight messages?
Also, what do you think about tacking on a time-out where one gives up on any outstanding messages and just sends the CLS?
|
closing this, it's been 6 years :) (issue #70 is still open tracking this) |
|
No worries. It was about time ;-) |
Attempting to handle issue #70 by building on @jonesetc close patch in issue #100 to fold in a CLS message back to the nsqd.
Close stuff happens in a lot of places, probably missed something.
I need to figure out how to roll in some tests as well.