lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Markus Jelsma <markus.jel...@openindex.io>
Subject RE: IndexFetcher cannot download index file
Date Tue, 24 Apr 2018 18:50:07 GMT
To be even more precise, it seems some WARN logs do show a stack trace in the GUI, but others
don't. For example:

org.apache.solr.common.SolrException: URLDecoder: Invalid character encoding detected after
position 23 of query string / form data (while parsing as UTF-8)
	at org.apache.solr.servlet.SolrRequestParsers.decodeChars(SolrRequestParsers.java:421)
	at org.apache.solr.servlet.SolrRequestParsers.decodeBuffer(SolrRequestParsers.java:437)
	at org.apache.solr.servlet.SolrRequestParsers.parseFormDataContent(SolrRequestParsers.java:404)
	at org.apache.solr.servlet.SolrRequestParsers.parseQueryString(SolrRequestParsers.java:304)

I don't know why there is a difference in showing stack traces.

 
 
-----Original message-----
> From:Markus Jelsma <markus.jelsma@openindex.io>
> Sent: Tuesday 24th April 2018 20:36
> To: solr-user@lucene.apache.org
> Subject: RE: IndexFetcher cannot download index file
> 
> Hello Shawn,
> 
> I should be more precise, i said the stack traces of WARN are not shown, only the messages
are visible. The 'low disk space' line was hidden in the stack trace of the WARN, as you can
see in the pasted example, thus invisible in the GUI with default settings.
> 
> If the log level of the message were raised to ERROR, it would be visible. I would think
that any recovery WARN message should be raised to ERROR or even FATAL, because in this state
whatever Solr does, it will never recover and in the mean time not show why the recovery failed.
> 
> What do you think?
> 
> Regards,
> Markus
> 
>  
>  
> -----Original message-----
> > From:Shawn Heisey <apache@elyograg.org>
> > Sent: Tuesday 24th April 2018 19:12
> > To: solr-user@lucene.apache.org
> > Subject: Re: IndexFetcher cannot download index file
> > 
> > On 4/24/2018 9:46 AM, Markus Jelsma wrote:
> > > Disk space was WARN level. It seems only stack traces of ERROR level messages
are visible via the GUI, and that is where the 'No space left' was hiding. Without logging
in and inspecting the logs manually, you will never notice that message.
> > 
> > The logging tab in the admin UI will by default show all messages at
> > WARN and higher severity.  If you're not seeing WARN messages there,
> > then a configuration has been changed to get rid of them.
> > 
> > Regarding general disk space requirements:  There should always be
> > enough disk space so that all the indexes can double in size
> > temporarily.  To be absolutely certain you won't run out, it should be
> > enough space so they can triple in size temporarily -- there is a
> > certain indexing scenario where this can happen in the wild. 
> > Replication can also create an entirely separate copy of the index
> > temporarily, so the same disk space recommendations apply for that too.
> > 
> > Thanks,
> > Shawn
> > 
> > 
> 

Mime
View raw message