This message was deleted.
# generals
Slackbot
08/07/2023, 11:38 PMThis message was deleted.
n
Neng
08/08/2023, 12:03 AMcc @Penghui
s
sijieg
08/08/2023, 1:48 AMhow do you cache the pulsar consumer? was the consumer closed when it is evicted out of the cache?
c
Chaoqin Li
08/08/2023, 1:52 AMI set a very large timeout to prevent cache eviction to exclude this possibility. I don’t think the consumer is closed.
Chaoqin Li
08/08/2023, 3:58 AMIt seems to be the expected behavior https://github.com/apache/pulsar/blob/2ab184e49a036a1dd10dc537bef4ab034a5ad5e0/pul[…]t/src/main/java/org/apache/pulsar/client/impl/ConsumerImpl.java
Chaoqin Li
08/08/2023, 4:55 AMIs there any config I can set in the broker side or client side to avoid running into this issue? Ideally we don't need to check whether reader or consumer is connected before seeking.
s
sijieg
08/08/2023, 7:08 AMIf you don't close the consumer proactively, I don't expect the consumer will disconnect itself unless there are some recent changes that I am not aware of. @Penghui thoughts?
p
Penghui
08/08/2023, 1:21 PM@sijieg It looks like only the seek method follows a different way. We should add a backoff policy here instead of throwing the above exception to the end user. cc @Tboy could you please help to make a fix?
internalGetLastMessageIdAsync👌 1
t
Tboy
08/09/2023, 2:48 AMPush a improvement for this : https://github.com/apache/pulsar/pull/20963
c
Chaoqin Li
08/11/2023, 6:56 AMBefore the fix is merged, is there any config I can set to mitigate this issue? For example, a broker config that make the connection sticky? @Penghui