alexku7 opened a new issue #8115:
URL: https://github.com/apache/pulsar/issues/8115
Hello
We experience the consumers stuck under very strange conditions. It’s still
hard to catch . It always happens with KEY_SHARED subscriptions with multiple
consumers.
What we figured out that the trigger for the problem is the adding
additional consumers. It can be occurred as part of upgrade ( for example the
new upgraded consumers being added and after this the old consumers being
removed from the topic) .
Sometimes , because of network glitch, consumer is disconnecting and after
few seconds reconnecting. We have 12 consumers on this particular topic but
it’s enough that only one consumer will disconnect and reconnect in order to
stuck the whole partition for every consumers while other partitions still
works fine.
Again, the common trigger is some change with consumers, preferably the
adding of new consumer/consumers
It doesn’t happen always. My estimation that it happens approximately in
half of cases.
Of course, no errors on the brokers side expecting the standard messages
about connecting new consumers.
The problem is that in order to recover from this situation we need to do
one of the following :
Pulsar-admin topics unload topicname
Or we need completely disconnect all consumers and reconnect ( it isn’t
recovering if even one consumers remaining connected.. we need to disconnect
all of them at once)
Any ideas ?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
