[
https://issues.apache.org/jira/browse/CASSANDRA-11840?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15300915#comment-15300915
]
Paulo Motta commented on CASSANDRA-11840:
-----------------------------------------
bq. 48 hours gives more time but isn't it too long for operator to figure out
if it's timed out or not?
The default value should be long enough to transfer a multi-GB file + rebuild
indexes, which can take several hours for very dense nodes with STCS (for
example, on CASSANDRA-11826 it took 1 hour to transfer a 1.5GB file, so it may
take 20 hours to transfer a 30GB sstable).
I think 24 hours is reasonable enough, given it has been infinite for a long
time due to CASSANDRA-11286, and after that was fixed we have been receiving
many reports of {{SocketTimeoutException}} since the current default of 1 hour
is not sufficient to transfer large files. On CASSANDRA-11841 with keep-alive
support we should be able to lower that down to 10 minutes or less.
Updated patch and resubmitted tests with new default of 24 hours.
> Set a more conservative default to streaming_socket_timeout_in_ms
> -----------------------------------------------------------------
>
> Key: CASSANDRA-11840
> URL: https://issues.apache.org/jira/browse/CASSANDRA-11840
> Project: Cassandra
> Issue Type: Sub-task
> Reporter: Paulo Motta
> Assignee: Paulo Motta
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
