[jira] [Commented] (SOLR-8248) Log a query as soon as it comes in and assign a unique id to it
[ https://issues.apache.org/jira/browse/SOLR-8248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15069907#comment-15069907 ] Shawn Heisey commented on SOLR-8248: SOLR-7887 is the issue to upgrade log4j. Feel free to join in there. > Log a query as soon as it comes in and assign a unique id to it > --- > > Key: SOLR-8248 > URL: https://issues.apache.org/jira/browse/SOLR-8248 > Project: Solr > Issue Type: Improvement > Components: Server >Affects Versions: 5.3 >Reporter: Pushkar Raste >Priority: Minor > > Often times when there is an OutOfMemory error Solr fails to log details > about query that might have caused it. Solr doesn't provide enough > information to investigate the root cause in such case. > We can log a query as soon as it comes in and reference it by it's unique id > to log details like Hits, Status and QTime when query finishes. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-8248) Log a query as soon as it comes in and assign a unique id to it
[ https://issues.apache.org/jira/browse/SOLR-8248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15069029#comment-15069029 ] Shawn Heisey commented on SOLR-8248: bq. Is there a reason Solr doesn't use log4j2 or logback as these are supposed to faster than log4j Boring TL;DR history: For most of its history, Solr has been using slf4j logging (from slf4j.org), bound to java.util.logging. In version 4.3, logging jars were removed from the war file, moved to jetty's lib/ext folder, and the binding was changed to log4j 1.2. I do not know why the older log4j was chosen, perhaps it was simply a familiar library. Speculation says that part of it may have come from loyalty to a fellow Apache project, and the fact that the library on slf4j.org does not support log4j2. Just this year, log4j 1.x was declared completely end of life, so we have an issue to upgrade the binding to log4j2 (still using slf4j within Solr), but this is not a simple drop-in replacement. Code changes will likely be required to keep the Logging tab in the admin UI working. Jar changes will also be required. > Log a query as soon as it comes in and assign a unique id to it > --- > > Key: SOLR-8248 > URL: https://issues.apache.org/jira/browse/SOLR-8248 > Project: Solr > Issue Type: Improvement > Components: Server >Affects Versions: 5.3 >Reporter: Pushkar Raste >Priority: Minor > > Often times when there is an OutOfMemory error Solr fails to log details > about query that might have caused it. Solr doesn't provide enough > information to investigate the root cause in such case. > We can log a query as soon as it comes in and reference it by it's unique id > to log details like Hits, Status and QTime when query finishes. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-8248) Log a query as soon as it comes in and assign a unique id to it
[ https://issues.apache.org/jira/browse/SOLR-8248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15068898#comment-15068898 ] Pushkar Raste commented on SOLR-8248: - [SOLR-6919|https://issues.apache.org/jira/browse/SOLR-6919] logs REST request, however request may get modified based on the solr config. [~yo...@apache.org] I do agree that correlating query would be a problem (especially in multi sharded envirnment), other alternative is to log query twice and control logging query the first time if a flag is turned on. This however would also increase disk footprint of log files a lot. Is there a reason Solr doesn't use log4j2 or logback as these are supposed to faster than log4j > Log a query as soon as it comes in and assign a unique id to it > --- > > Key: SOLR-8248 > URL: https://issues.apache.org/jira/browse/SOLR-8248 > Project: Solr > Issue Type: Improvement > Components: Server >Affects Versions: 5.3 >Reporter: Pushkar Raste >Priority: Minor > > Often times when there is an OutOfMemory error Solr fails to log details > about query that might have caused it. Solr doesn't provide enough > information to investigate the root cause in such case. > We can log a query as soon as it comes in and reference it by it's unique id > to log details like Hits, Status and QTime when query finishes. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-8248) Log a query as soon as it comes in and assign a unique id to it
[ https://issues.apache.org/jira/browse/SOLR-8248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14995675#comment-14995675 ] Mike Drob commented on SOLR-8248: - Does SOLR-6919 get you most of the way there? > Log a query as soon as it comes in and assign a unique id to it > --- > > Key: SOLR-8248 > URL: https://issues.apache.org/jira/browse/SOLR-8248 > Project: Solr > Issue Type: Improvement > Components: Server >Affects Versions: 5.3 >Reporter: Pushkar Raste >Priority: Minor > > Often times when there is an OutOfMemory error Solr fails to log details > about query that might have caused it. Solr doesn't provide enough > information to investigate the root cause in such case. > We can log a query as soon as it comes in and reference it by it's unique id > to log details like Hits, Status and QTime when query finishes. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-8248) Log a query as soon as it comes in and assign a unique id to it
[
https://issues.apache.org/jira/browse/SOLR-8248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14995408#comment-14995408
]
Yonik Seeley commented on SOLR-8248:
It's definitely useful for debugging... I added something similar for debugging
updates in LogUpdateProcessor:
if (logDebug) { log.debug("PRE_UPDATE " + cmd.toString() + " " + req); }
But logging is certainly costly, so it probably shouldn't be enabled by
default. It would also make life slightly harder for basic log analysis by
requiring correlation by the id.
> Log a query as soon as it comes in and assign a unique id to it
> ---
>
> Key: SOLR-8248
> URL: https://issues.apache.org/jira/browse/SOLR-8248
> Project: Solr
> Issue Type: Improvement
> Components: Server
>Affects Versions: 5.3
>Reporter: Pushkar Raste
>Priority: Minor
>
> Often times when there is an OutOfMemory error Solr fails to log details
> about query that might have caused it. Solr doesn't provide enough
> information to investigate the root cause in such case.
> We can log a query as soon as it comes in and reference it by it's unique id
> to log details like Hits, Status and QTime when query finishes.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-8248) Log a query as soon as it comes in and assign a unique id to it
[ https://issues.apache.org/jira/browse/SOLR-8248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14995400#comment-14995400 ] Jason Gerlowski commented on SOLR-8248: --- >From a novice's perspective, this seems like a useful improvement, +1. Though >maybe there's performance-issues, or other things preventing this that I don't >know about. > Log a query as soon as it comes in and assign a unique id to it > --- > > Key: SOLR-8248 > URL: https://issues.apache.org/jira/browse/SOLR-8248 > Project: Solr > Issue Type: Improvement > Components: Server >Affects Versions: 5.3 >Reporter: Pushkar Raste >Priority: Minor > > Often times when there is an OutOfMemory error Solr fails to log details > about query that might have caused it. Solr doesn't provide enough > information to investigate the root cause in such case. > We can log a query as soon as it comes in and reference it by it's unique id > to log details like Hits, Status and QTime when query finishes. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
