Joseph Lynch commented on CASSANDRA-14214:

[~krummas] do you think it would be useful to have this as a compaction param 
(e.g. {{never_purge_tombstones }} similar to 
{{unchecked_tombstone_compaction}})? For example, if a user wants to use this 
option to rescue a single machine this patch is awesome, but if they want to 
set it on the entire cluster just for that particular CF a compaction parameter 
may be easier? If you think this is a good idea I can take a whack at a patch 
and submit it as a separate jira. This option has been a lifesaver for us for 
clusters when compaction falls behind because we can get 100-1000x faster 
compactions, rescue the cluster, and then disable it again.

Separately I was thinking that for TWCS or other "write once with TTL" column 
families it may be good to set {{never_purge_tombstones}} permanently if the 
user already has {{ignore_overlapping}} set (via 
{{unsafe_aggressive_sstable_expiration}}). Namely I don't think we need the 
short circuit on  [line 176 
 if {{ignoreOverlaps}} is true right?

> Make it possible to change NEVER_PURGE_TOMBSTONES during runtime
> ----------------------------------------------------------------
>                 Key: CASSANDRA-14214
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-14214
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Marcus Eriksson
>            Assignee: Marcus Eriksson
>            Priority: Major
>             Fix For: 4.x
> If a node gets behind on compaction it could be good to be able to flip the 
> NEVER_PURGE_TOMBSTONES flag to speed up compactions during runtime.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

Reply via email to