[
https://issues.apache.org/jira/browse/FLUME-109?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ashish Paliwal resolved FLUME-109.
----------------------------------
Resolution: Won't Fix
Fix Version/s: v0.9.5
Won't fix. 0.X branch not maintained anymore
> Node doesn't auto restart after HD full recovery
> ------------------------------------------------
>
> Key: FLUME-109
> URL: https://issues.apache.org/jira/browse/FLUME-109
> Project: Flume
> Issue Type: Bug
> Components: Node
> Affects Versions: v0.9.4
> Reporter: Nicholas Verbeck
> Fix For: v0.9.5
>
>
> If you have a node thats HD get full and you recover from this state. The
> node seems to hang and must be restarted before it will start sending data to
> the collectors. Ideally the node should detect the HD being full and monitor
> it tell it gets room again and auto start back up. It may also be worth while
> to have a config that is used for a HD available space threshold. That causes
> the processes to wait tell space becomes available.
> The issue that caused me to find this problem is that it also appears the
> write ahead logic seems to not regulate itself on how much data it writes to
> the "sent" folder before waiting for it to actually be sent and received. So
> a 25MB file will be expanded to ~2.5GB of seq (6582) files in the "sent"
> folder. This is assuming that the sent folder is being used as a buffer for
> the write ahead logic before actually sending the data.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
