[
https://issues.apache.org/jira/browse/ACCUMULO-1410?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13762151#comment-13762151
]
ASF subversion and git services commented on ACCUMULO-1410:
-----------------------------------------------------------
Commit 5d7a4a727247d7d19ea29b1031bedcefb690b4a1 in branch refs/heads/master
from [[email protected]]
[ https://git-wip-us.apache.org/repos/asf?p=accumulo.git;h=5d7a4a7 ]
ACCUMULO-1410 - making the ZK connections a bit tolerant to user requests
> ZooSession.connect barely adheres to timeout
> --------------------------------------------
>
> Key: ACCUMULO-1410
> URL: https://issues.apache.org/jira/browse/ACCUMULO-1410
> Project: Accumulo
> Issue Type: Bug
> Components: client, fate
> Reporter: John Vines
> Assignee: John Vines
> Fix For: 1.6.0
>
> Attachments: ACCUMULO-1410.patch
>
>
> ZooSession.connect, which is used by ZooKeeperInstance, takes an argument for
> a timeout, and utilizes it to an extent-
> {quote}
> if (System.currentTimeMillis() - startTime > 2 * timeout)
> {quote}
> However, this is only used after a check which uses hardcoded values.
> Currently, this is set to 10*1000ms. More specifically, it uses this value
> and checks every 100ms to see if it's connected. So if you have a tiny
> timeout, there are 2 issues:
> # Your timeout is only useful in 10 second increments, rounded up
> # You get a nice helpful error message that hides that real lengths of attempt
> I think the block of code should be changed to just try to connect for the
> user specified timeout length, working in the same 100ms increments. This
> allows more granularity in the handling of the user specified values (and I
> think it also simplifies the code). This will also make the timeout message
> more accurate.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
