Better warning message on cluster sharding registration #24906
Conversation
akka-ci
added
validating
|
Test PASSed. |
| coordinatorSelection, shardBuffers.totalSize) | ||
| actorSelection, shardBuffers.totalSize) | ||
| case None ⇒ log.warning( | ||
| "No coordinator found to register. Probably, no seed-nodes configured and manual cluster join not performed? Total [{}] buffered messages.", |
There was a problem hiding this comment.
This can also happen when members are joining but not yet Up - perhaps we could find a simple way to word this that makes that more obvious?
There was a problem hiding this comment.
Below cases are possible for coordinatorSelection being None ?
- seed-node not configured and manual configuration is not performed
- members are joining but not yet
Up - general network problem
I think the original issue #24295 intended to make 1. more obvious, and distinguished from the case where coordinatorSelection is Some(). However, I'm not sure how to put these 1 ~ 3 cases into a single warning message with reasonable length... so, what I think could do are:
- put a lengthy message saying 1, 2, and 3 are possible
- or, link to a certain section of the doc explaining 1, 2 and 3
There was a problem hiding this comment.
I think this distinction between None and Some makes it much more clear what the reason might be. Not Up yet should be rare, since coordinator is on the oldest.
One additional hint to the log message of the Some case could be to include something if it's unreachable:
val coordinatorIsUnreachable = cluster.state.unreachable(membersByAge.head)
| coordinatorSelection, shardBuffers.totalSize) | ||
| actorSelection, shardBuffers.totalSize) | ||
| case None ⇒ log.warning( | ||
| "No coordinator found to register. Probably, no seed-nodes configured and manual cluster join not performed? Total [{}] buffered messages.", |
There was a problem hiding this comment.
I think this distinction between None and Some makes it much more clear what the reason might be. Not Up yet should be rare, since coordinator is on the oldest.
One additional hint to the log message of the Some case could be to include something if it's unreachable:
val coordinatorIsUnreachable = cluster.state.unreachable(membersByAge.head)
akka-ci
added
validating
|
Test PASSed. |
akka-ci
added
validating
richardimaoka
force-pushed
the
wip-24295-cluster-sharding-error-richard
branch
from
d45cc70
to
077ed2f
Compare
|
Added a new commit to this - @patriknw does this look ok? |
akka-ci
added
validating
|
Test FAILed. |
|
argh, will look into the test failure |
akka-ci
added
tested
|
Test PASSed. |
|
Test passed, and ready for review now. The earlier test failure was due to a wrong character I accidentally inserted in a comment, which I removed by git commit --amend. |
* Provide access to known shard types akka/akka#23912 * Separate sharding regions and proxies akka/akka#23472 Fix lookup of coordinator for sharding proxies akka/akka#23995 * Fix race in ClusterShardingFailureSpec AFAICT there was nothing ensuring the order of messages when sent to the shard and the region so first checkthat the passivation has happened before sending another add in the test akka/akka#24013 * Better warning message on cluster sharding registration akka/akka#24906 * entityId => Behavior in ClusterSharding API mixture of akka/akka#24053 akka/akka#21809 akka/akka#24470 * sharding tests updated * headers fixed, docs updated * ClusterSharding: automatically choose start or startProxy by a node role akka/akka#23934
Closes #24295