You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@activemq.apache.org by "Gary Tully (JIRA)" <ji...@apache.org> on 2018/03/02 10:27:00 UTC
[jira] [Comment Edited] (AMQ-4261) Please change logging message
from INFO to WARN or ERROR when persistent store is full
[ https://issues.apache.org/jira/browse/AMQ-4261?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16383429#comment-16383429 ]
Gary Tully edited comment on AMQ-4261 at 3/2/18 10:26 AM:
----------------------------------------------------------
I am coming around to the view that this should be a WARN, b/c it typically impacts the application and is indicative of slow consumers which can often be a problem.
It is info for sure in a development env, but in a production env where the WARN threshold is typical, it is good information that the broker can provide to an operator.
In addition, we now have the policyEntry blockedProducerWarningInterval which controls how often this event is emitted which allows operators to limit the impact on logging if this behaviour is the norm and creates unnecessary noise. blockedProducerWarningInterval=0 will disable it and there will just be debug level logging on this event.
was (Author: gtully):
I am coming around to the view that this should be a WARN, b/c it typically impacts the application and is indicative of slow consumers which can often be a problem.
It is info for sure for in a development env, but in a production env where WARN is typical is is good information that the broker can provide to an operator to WARN them.
In addition, we now have the policyEntry blockedProducerWarningInterval which controls how often this event is emitted which allows operators to limit the impact on logging if this behaviour is the norm and creates unnecessary noise. blockedProducerWarningInterval=0 will disable it and there will just be debug level logging on this event.
> Please change logging message from INFO to WARN or ERROR when persistent store is full
> ---------------------------------------------------------------------------------------
>
> Key: AMQ-4261
> URL: https://issues.apache.org/jira/browse/AMQ-4261
> Project: ActiveMQ
> Issue Type: Improvement
> Affects Versions: 5.5.1
> Reporter: Vincent Gao
> Assignee: Gary Tully
> Priority: Major
>
> We use KahaDB for ActiveMQ persistence. When queue's persistent store is full, a message such as the following is logged at the INFO level.
> We'd like to request the message to be logged at the ERROR or WARN level. This is because production systems typically configures loggers to display messages at WARN level. A situation such as this gets into deadlock and makes the application unusable. Displaying the message at the ERROR or WARN
> level will provide visibility about the issue for typical production usage.
> 2013-01-17 10:56:28,563 | INFO | iveMQ Task-37644 | Queue | 118 - org.apache.activemq.activemq-core - 5.5.1.fuse-7-061 | Usage
> (default:store:queue://xrm-BundleStagingQ:store) percentUsage=99%, usage=1094475857, limit=1073741824, percentUsageMinDelta=1%;Parent:Usage(default:store)
> percentUsage=101%, usage=1094475857, limit=1073741824, percentUsageMinDelta=1%: Persistent store is Full, 100% of 1073741824. Stopping producer (ID:vlcov
> 46.fmr.com-47345-1358284999111-4:6:2548:1) to prevent flooding queue://xrm-BundleStagingQ. See http://activemq.apache.org/producer-flow-control.html for m
> ore info (blocking for: 29828s)
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)