hozumit opened a new issue #11920:
URL: https://github.com/apache/pulsar/issues/11920


   **Describe the bug**
   I found that sometimes a topic remains unassigned after unloading.
   The following message is repeatedly appeared in broker log.
   
   ```
   08:28:28.538 [pulsar-io-4-15] WARN  
org.apache.pulsar.broker.service.BrokerService - Namespace is being unloaded, 
cannot add topic persistent://mytenant/mynamespace/xxx-partition-37
   ```
   
   The following log is the beginning of unloading.
   ```
   09:57:13.634 [pulsar-web-40-7] INFO  
org.apache.pulsar.broker.admin.impl.NamespacesBase - [null] Unloading namespace 
bundle mytenant/mynamespace/0x7ef23d0b_0x80000000
   09:57:13.725 [pulsar-web-40-7] INFO  
org.apache.pulsar.broker.namespace.OwnedBundle - Disabling ownership: 
mytenant/mynamespace/0x7ef23d0b_0x80000000
   09:57:13.727 [pulsar-web-40-7] INFO  
org.apache.pulsar.broker.service.BrokerService - 
[persistent://mytenant/mynamespace/xxx-partition-37] Unloading topic
   ```
   
   ```
   $ docker exec -it pulsar_broker bin/pulsar-admin topics stats 
"persistent://mytenant/mynamespace/xxx-partition-37"
   Can't find owner for topic persistent://mytenant/mynamespace/xxx-partition-37
   
   Reason: Can't find owner for topic 
persistent://mytenant/mynamespace/xxx-partition-37
   ```
   
   I saw this issue several times on different partitions of a topic, which 
have heavy loads.
   I have solved the problem by restarting a broker process(docker container).
   
   **To Reproduce**
   Steps to reproduce the behavior:
   Trigger unloading by command line( `bin/pulsar-admin namespaces unload 
mytenant/mynamespace` ) or broker load shedding.
   Topic unloading and assignment mostly works. I think that it fails when 
brokers are in relatively heavy loads.
   
   **Expected behavior**
   I don't know if the allocation failure is a bug or not, but brokers should 
at least try to re-allocate the topic as soon as possible after the allocation 
failure.
   I have solved the problem by restarting the broker process, so I think 
successive allocations will work.
   
   **Desktop (please complete the following information):**
    - OS: CentOS
    - Pulsar (broker and client): v2.8.0
   
   **Additional context**
   


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to