You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by binit <b....@gmail.com> on 2013/12/13 05:48:14 UTC
Unable to check Solr 4.6 SPLITSHARD command progress
I have a big index, approx size 350 GB in a single shard, which I want to
split.
The SPLITSHARD command initiates successfully as I can see in the logs. (It
times out but reading the forums here it is the expected behavior).
The problem is it never completes even after a full day, and I doubt it is
actually running in the background. Because the created two shard folders
are of constant size in KBs and is not changing over time.
In the zookeeper the two shards are active and marked as under construction.
The logs do not show any error, the last log entry is like:
INFO - 2013-12-12 14:48:36.110; org.apache.solr.update.SolrIndexSplitter;
SolrIndexSplitter: partition #0 range=80000000-ffffffff
Now, what I'm asking is if it is possible to check the progress of this
command. Even if it is as simple as changing of files in some directory...
Or taking dump of threads, which just assures me that it is running &
progressing.
I'd also be happy to know how does the overseer split. I also tried looking
into its code but was unable to figure out the way it does so. Also, in the
zookeeper files related to overseer I was unable to find anything related to
the split command.
Thanks,
Binit
--
View this message in context: http://lucene.472066.n3.nabble.com/Unable-to-check-Solr-4-6-SPLITSHARD-command-progress-tp4106520.html
Sent from the Solr - User mailing list archive at Nabble.com.
RE: Unable to check Solr 4.6 SPLITSHARD command progress
Posted by Craig Christman - US <cc...@caci.com>.
We ran into this exact scenario and resolved by applying SOLR-5214 (://issues.apache.org/jira/browse/SOLR-5214)
________________________________________
From: binit [b.initthapa@gmail.com]
Sent: Friday, December 13, 2013 10:45 PM
To: solr-user@lucene.apache.org
Subject: Re: Unable to check Solr 4.6 SPLITSHARD command progress
Yes, and my clusterstate.json is still:
==
"shards":{
"shard1":{
"range":"80000000-7fffffff",
"state":"active",
"replicas":{"core_node1":{
"state":"active",
"base_url":"http://...../solr",
"core":".._shard1_replica1",
"node_name":"....:8080_solr",
"leader":"true"}}},
"shard1_1":{
"range":"0-7fffffff",
"state":"construction",
"parent":"shard1",
"replicas":{"core_node2":{
"state":"active",
"base_url":"http://....:8080/solr",
"core":".._shard1_1_replica1",
"node_name":"....:8080_solr",
"leader":"true"}}},
"shard1_0":{
"range":"80000000-ffffffff",
"state":"construction",
"parent":"shard1",
"replicas":{"core_node3":{
"state":"active",
"base_url":"http://....:8080/solr",
"core":".._shard1_0_replica1",
"node_name":"....:8080_solr",
"leader":"true"}}}},
"maxShardsPerNode":"1",
"router":{"name":"compositeId"},
"replicationFactor":"1"}}
==
But, it got failed finally with out of memory.
And is definitely not progressing because the thread is stopped. Probably
SPLITSHARD is not mature enough to use yet.
Now, I've no choice but to do it from Solrj, indexing manually.
--
View this message in context: http://lucene.472066.n3.nabble.com/Unable-to-check-Solr-4-6-SPLITSHARD-command-progress-tp4106520p4106699.html
Sent from the Solr - User mailing list archive at Nabble.com.
Re: Unable to check Solr 4.6 SPLITSHARD command progress
Posted by binit <b....@gmail.com>.
Yes, and my clusterstate.json is still:
==
"shards":{
"shard1":{
"range":"80000000-7fffffff",
"state":"active",
"replicas":{"core_node1":{
"state":"active",
"base_url":"http://...../solr",
"core":".._shard1_replica1",
"node_name":"....:8080_solr",
"leader":"true"}}},
"shard1_1":{
"range":"0-7fffffff",
"state":"construction",
"parent":"shard1",
"replicas":{"core_node2":{
"state":"active",
"base_url":"http://....:8080/solr",
"core":".._shard1_1_replica1",
"node_name":"....:8080_solr",
"leader":"true"}}},
"shard1_0":{
"range":"80000000-ffffffff",
"state":"construction",
"parent":"shard1",
"replicas":{"core_node3":{
"state":"active",
"base_url":"http://....:8080/solr",
"core":".._shard1_0_replica1",
"node_name":"....:8080_solr",
"leader":"true"}}}},
"maxShardsPerNode":"1",
"router":{"name":"compositeId"},
"replicationFactor":"1"}}
==
But, it got failed finally with out of memory.
And is definitely not progressing because the thread is stopped. Probably
SPLITSHARD is not mature enough to use yet.
Now, I've no choice but to do it from Solrj, indexing manually.
--
View this message in context: http://lucene.472066.n3.nabble.com/Unable-to-check-Solr-4-6-SPLITSHARD-command-progress-tp4106520p4106699.html
Sent from the Solr - User mailing list archive at Nabble.com.
Re: Unable to check Solr 4.6 SPLITSHARD command progress
Posted by Yago Riveiro <ya...@gmail.com>.
You can see if the SPLITSHARD finished in clusterstate.json, if the original shard has his state as inactive and the children active, the operation finished satisfactorily.
--
Yago Riveiro
Sent with Sparrow (http://www.sparrowmailapp.com/?sig)
On Friday, December 13, 2013 at 4:48 AM, binit wrote:
> I have a big index, approx size 350 GB in a single shard, which I want to
> split.
>
> The SPLITSHARD command initiates successfully as I can see in the logs. (It
> times out but reading the forums here it is the expected behavior).
>
> The problem is it never completes even after a full day, and I doubt it is
> actually running in the background. Because the created two shard folders
> are of constant size in KBs and is not changing over time.
>
> In the zookeeper the two shards are active and marked as under construction.
> The logs do not show any error, the last log entry is like:
> INFO - 2013-12-12 14:48:36.110; org.apache.solr.update.SolrIndexSplitter;
> SolrIndexSplitter: partition #0 range=80000000-ffffffff
>
> Now, what I'm asking is if it is possible to check the progress of this
> command. Even if it is as simple as changing of files in some directory...
> Or taking dump of threads, which just assures me that it is running &
> progressing.
>
> I'd also be happy to know how does the overseer split. I also tried looking
> into its code but was unable to figure out the way it does so. Also, in the
> zookeeper files related to overseer I was unable to find anything related to
> the split command.
>
> Thanks,
> Binit
>
>
>
> --
> View this message in context: http://lucene.472066.n3.nabble.com/Unable-to-check-Solr-4-6-SPLITSHARD-command-progress-tp4106520.html
> Sent from the Solr - User mailing list archive at Nabble.com (http://Nabble.com).
>
>