lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Shawn Heisey <apa...@elyograg.org>
Subject Re: Could not load collection from ZK:
Date Tue, 20 Jun 2017 12:37:30 GMT
On 6/20/2017 6:08 AM, Aman Deep Singh wrote:
> I'm facing a issue in solr sometimes zookeeper failes to load the solr
> collection stating
>
> org.apache.solr.common.SolrException: Could not load collection from ZK:

This is not the full error message.  It will be dozens of lines long,
and may contain one or more "Caused by" sections with separate
stacktraces.  We need the whole thing, with complete text of all lines
included.  If you are looking at the admin UI to see these messages, you
need to look in the solr.log file instead.  It looks like you've used
the service installer script, so you should find that in /var/solr/logs.

> My current setup details is
>
>    1. 5 Nodes  with 4 cores  ,7.6 GB RAM each which contains solr node and
>    zookeeper
>    2. No sharding is used
>    3. index size is around 2.5 GB
>    4. solrnode RAM-4GB
>    5. Average load -10k RPM
>    6. Indexing maximum 1000 docs/Minute  (around 100 batch of 5 to 10 docs
>    each)
>
> The GC logs analysis is also normal
> Node 1-
> http://gceasy.io/my-gc-report.jsp?p=c2hhcmVkLzIwMTcvMDYvMjAvLS1zb2xyX2djLmxvZy4yLnppcC0tMTEtNTctMw==
> Node 2-
> http://gceasy.io/my-gc-report.jsp?p=c2hhcmVkLzIwMTcvMDYvMjAvLS1zb2xyX2djLmxvZy4zLmN1cnJlbnQuemlwLS03LTQ2LTU2
> Node 3-
> http://gceasy.io/my-gc-report.jsp?p=c2hhcmVkLzIwMTcvMDYvMjAvLS1zb2xyX2djLmxvZy4zLmN1cnJlbnQuemlwLS04LTIzLTM5
> Node 4-
> http://gceasy.io/my-gc-report.jsp?p=c2hhcmVkLzIwMTcvMDYvMjAvLS1zb2xyX2djLmxvZy4yLnppcC0tOC0yMC01NQ==
> Node 5-
> http://gceasy.io/my-gc-report.jsp?p=c2hhcmVkLzIwMTcvMDYvMjAvLS1zb2xyX2djLmxvZy41LmN1cnJlbnQuemlwLS04LTE5LTE0

Overall the GC stats look good.  There are a few long GCs, but the "stop
the world" stats are pretty good.

> Admin UI image https://pasteboard.co/1Nd7ArAf0.png

This screenshot indicates that this system is using swap.  This is
usually an indication that there's an overall memory shortage on the
system ... but if there is no *active* swapping, there's a *small*
chance that this is not a problem.  What I would like to see is a
screenshot with the "top" program running (not htop or any other
variant, I'm after "top" itself), with shift-M pressed to sort the list
by memory usage.

Thanks,
Shawn


Mime
View raw message