[
https://issues.apache.org/jira/browse/CLOUDSTACK-8325?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14359864#comment-14359864
]
Abhinandan Prateek commented on CLOUDSTACK-8325:
------------------------------------------------
On further investigation:
1. If we use date extension then log rotate then logs are rotated once daily as
has been set.
2. If we try to force log rotate or do a hourly cron, the logs won;t rotate as
the "date" extension is already taken.
3. If some process is writing into logs when log rotate runs, it is possible
that compression fails as "name change" to gz will not happen. If that breaks
logger continues writing to the last log file, the process is broken.
Here is what I suggest:
1. Drop date extension
2. Run log rotate in hourly cron
3. delay compress (for 3).
> it is observed that the VR fails as the log files have taken up all the disk
> -----------------------------------------------------------------------------
>
> Key: CLOUDSTACK-8325
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8325
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Components: Virtual Router
> Affects Versions: 4.3.2
> Reporter: Abhinandan Prateek
> Assignee: Abhinandan Prateek
> Priority: Critical
> Fix For: 4.6.0
>
>
> It is observed from support cases logged that the log files sometime grow so
> fast that they take away all the available disk space on the VR.
> It could happen because 1. Either log rotate did not happen in timely fashion
> 2. There was intensive logging due to repeated changes in firewall rules (or
> API calls).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
