lhotari commented on PR #22597:
URL: https://github.com/apache/pulsar/pull/22597#issuecomment-2081992936
It seems that __change_events has special handling in replication:
lhotari commented on PR #22597:
URL: https://github.com/apache/pulsar/pull/22597#issuecomment-2081936396
Same exception with current master branch version 93afd89b047
```
2024-04-29T08:47:36,525 - WARN - [pulsar-io-35-4:ServerCnx] - Failed to get
Partitioned Metadata
heesung-sn commented on PR #22597:
URL: https://github.com/apache/pulsar/pull/22597#issuecomment-2078763715
To answer your questions, Im not sure. I assume all system topics should
live in local cluster only,but I could be wrong.
--
This is an automated message from the Apache Git
heesung-sn commented on PR #22597:
URL: https://github.com/apache/pulsar/pull/22597#issuecomment-2078761854
I realized that Pulsar has some loopholes that unnecessarily try to
replicate, migrate or do other operations on system topics.
We need to check all system topics and define
lhotari commented on PR #22597:
URL: https://github.com/apache/pulsar/pull/22597#issuecomment-2078730928
@poorbarcode @heesung-sn Do you think that relaxing the
checkLocalOrGetPeerReplicationCluster checks for system topics would be useful?
--
This is an automated message from the
lhotari opened a new pull request, #22597:
URL: https://github.com/apache/pulsar/pull/22597
WIP, not ready for review. PR opened in draft mode for initial discussion
and feedback.
### Motivation
When investigating another issue and reproducing the problem using these