You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@nifi.apache.org by "Joseph Witt (JIRA)" <ji...@apache.org> on 2015/08/25 06:32:45 UTC
[jira] [Commented] (NIFI-893) If FlowFile Repository runs out of
disk space, NiFi stops processing until restarted
[ https://issues.apache.org/jira/browse/NIFI-893?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14710570#comment-14710570 ]
Joseph Witt commented on NIFI-893:
----------------------------------
[~markap14] Can you describe a bit about why you see this as an 0.4.0 thing? In general we have always expected and should always expect that disk space will be exhausted. In such cases nifi should 'do the right thing' such that it won't do anything to cause data loss or unsafe operation but that once a given condition is resolved (somehow) then NiFi will resume operation. I understand that if the flow file repo runs out of space then there really isn't much you can do nor any real reason why to assume it would work itself out. If the flow file repo is on its own partition and fills up then yeah I can see how not much could be done. But if it is on a repo sharing space with other repo or things on the os/filesystem then it is reasonable to assume it will be a temporary condition. So we should be resilient to such cases independent of a manual restart.
> If FlowFile Repository runs out of disk space, NiFi stops processing until restarted
> ------------------------------------------------------------------------------------
>
> Key: NIFI-893
> URL: https://issues.apache.org/jira/browse/NIFI-893
> Project: Apache NiFi
> Issue Type: Bug
> Components: Core Framework
> Reporter: Mark Payne
> Fix For: 0.4.0
>
>
> If the FlowFile Repository runs out of disk space, NiFi stops processing any data. Once space is recovered on the partition, NiFi never resumes processing until it is restarted. A restart appears to have recovered all data without issue and all appears to work as expected after restarting.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)