You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@pinot.apache.org by "itschrispeck (via GitHub)" <gi...@apache.org> on 2023/07/17 17:24:41 UTC
[GitHub] [pinot] itschrispeck commented on issue #11004: Follower server redownloads segments upon server restart due to CRC change
itschrispeck commented on issue #11004:
URL: https://github.com/apache/pinot/issues/11004#issuecomment-1638567703
You're right, I think it's coincidence that the clusters we saw issue with are running a newer version of Pinot. We've jumped ahead to a commit that includes both fixes and we're still seeing the behavior.
It looks like the lucene index may not be created deterministically also:
```
Binary files replica_1/json_data.lucene.index/_0.cfe and replica_2/json_data.lucene.index/_0.cfe differ
72 | Binary files replica_1/json_data.lucene.index/_0.cfs and replica_2/json_data.lucene.index/_0.cfs differ
73 | Binary files replica_1/json_data.lucene.index/_0.si and replica_2/json_data.lucene.index/_0.si differ
74 | Binary files replica_1/json_data.lucene.index/segments_1 and replica_2/json_data.lucene.index/segments_1 differ
75 | Binary files replica_1/message_dictionaryVars.lucene.index/_0.cfe and replica_2/message_dictionaryVars.lucene.index/_0.cfe differ
76 | Binary files replica_1/message_dictionaryVars.lucene.index/_0.cfs and replica_2/message_dictionaryVars.lucene.index/_0.cfs differ
77 | Binary files replica_1/message_dictionaryVars.lucene.index/_0.si and replica_2/message_dictionaryVars.lucene.index/_0.si differ
78 | Binary files replica_1/message_dictionaryVars.lucene.index/segments_1 and replica_2/message_dictionaryVars.lucene.index/segments_1 differ
79 | Binary files replica_1/message_logtype.lucene.index/_0.cfe and replica_2/message_logtype.lucene.index/_0.cfe differ
80 | Binary files replica_1/message_logtype.lucene.index/_0.cfs and replica_2/message_logtype.lucene.index/_0.cfs differ
81 | Binary files replica_1/message_logtype.lucene.index/_0.si and replica_2/message_logtype.lucene.index/_0.si differ
82 | Binary files replica_1/message_logtype.lucene.index/segments_1 and replica_2/message_logtype.lucene.index/segments_1 differ
83
```
And I think this issue would also present if we use certain transform functions (e.g. storing ingestion time via `now()`).
What necessitates computing the CRC for all segments each restart? Assuming no data corruption happened, it seems that any data difference was already being 'served' as valid data.
--
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: commits-unsubscribe@pinot.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@pinot.apache.org
For additional commands, e-mail: commits-help@pinot.apache.org