[
https://issues.apache.org/jira/browse/SOLR-11807?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Varun Thacker updated SOLR-11807:
---------------------------------
Description:
When you start Solr 6.6. and run the cloud example here's the log excerpt :
{code:java}
INFO - 2018-06-20 13:44:47.452;
org.apache.solr.client.solrj.impl.ZkClientClusterStateProvider; Cluster at
localhost:9983 ready
Connecting to ZooKeeper at localhost:9983 ...
INFO - 2018-06-20 13:44:47.491;
org.apache.solr.client.solrj.impl.ZkClientClusterStateProvider; Cluster at
localhost:9983 ready
Uploading
/Users/varunthacker/Downloads/search-downloads/solr-6.6.0/server/solr/configsets/data_driven_schema_configs/conf
for config gettingstarted to ZooKeeper at localhost:9983
Creating new collection 'gettingstarted' using command:
http://localhost:8983/solr/admin/collections?action=CREATE&name=gettingstarted&numShards=2&replicationFactor=2&maxShardsPerNode=2&collection.configName=gettingstarted{code}
maxShardsPerNode get's set to 2 .
Compare this to Solr 7.3
{code:java}
INFO - 2018-06-20 13:55:33.823;
org.apache.solr.client.solrj.impl.ZkClientClusterStateProvider; Cluster at
localhost:9983 ready
Created collection 'gettingstarted' with 2 shard(s), 2 replica(s) with
config-set 'gettingstarted'{code}
So something changed and now we no longer set maxShardsPerNode and it defaults
to -1 .
-1 has special handing while creating a collection ( it means max int ) . This
special handling is not there while Restoring a collection and hence this fails
We should not set maxShardsPerNode to -1 in the first place
Steps to reproduce:
# ./bin/solr start -e cloud -noprompt : This creates a 2 node cluster and a
gettingstarted collection which 2X2
2. Added 4 docs (id=1,2,3,4) with commit=true and openSearcher=true (default)
3. Call backup:
http://localhost:8983/solr/admin/collections?action=BACKUP&name=gettingstarted_backup&collection=gettingstarted&location=/Users/varunthacker/solr-7.1.0
4. Call restore:
[http://localhost:8983/solr/admin/collections?action=restore&name=gettingstarted_backup&collection=restore_gettingstarted&location=/Users/varunthacker/solr-7.1.0]
was:
Today we can never restore the cloud example gettingstarted collection because
it sets maxShardsPerNode=-1 . We have special handling for -1 in
CreateCollectionCmd but not in RestoreCmd.
Today RestoreCmd does has this check and it fails to restore the collection
{code}
if ((numShards * totalReplicasPerShard) > (availableNodeCount *
maxShardsPerNode)) {
throw new SolrException(ErrorCode.BAD_REQUEST,
String.format(Locale.ROOT, "Solr cloud with available number of
nodes:%d is insufficient for"
+ " restoring a collection with %d shards, total replicas per
shard %d and maxShardsPerNode %d."
+ " Consider increasing maxShardsPerNode value OR number of
available nodes.",
availableNodeCount, numShards, totalReplicasPerShard,
maxShardsPerNode));
}
{code}
Steps to reproduce:
Steps to reproduce:
1. ./bin/solr start -e cloud -noprompt : This creates a 2 node cluster and a
gettingstarted collection which 2X2
2. Added 4 docs (id=1,2,3,4) with commit=true and openSearcher=true (default)
3. Call backup:
http://localhost:8983/solr/admin/collections?action=BACKUP&name=gettingstarted_backup&collection=gettingstarted&location=/Users/varunthacker/solr-7.1.0
4. Call restore:
http://localhost:8983/solr/admin/collections?action=restore&name=gettingstarted_backup&collection=restore_gettingstarted&location=/Users/varunthacker/solr-7.1.0
> Restore Collection should reuse maxShardsPerNode specified during backup
> ------------------------------------------------------------------------
>
> Key: SOLR-11807
> URL: https://issues.apache.org/jira/browse/SOLR-11807
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: Backup/Restore
> Reporter: Varun Thacker
> Assignee: Varun Thacker
> Priority: Minor
> Fix For: master (8.0), 7.5
>
> Attachments: SOLR-11807.patch, SOLR-11807.patch
>
>
> When you start Solr 6.6. and run the cloud example here's the log excerpt :
> {code:java}
> INFO - 2018-06-20 13:44:47.452;
> org.apache.solr.client.solrj.impl.ZkClientClusterStateProvider; Cluster at
> localhost:9983 ready
> Connecting to ZooKeeper at localhost:9983 ...
> INFO - 2018-06-20 13:44:47.491;
> org.apache.solr.client.solrj.impl.ZkClientClusterStateProvider; Cluster at
> localhost:9983 ready
> Uploading
> /Users/varunthacker/Downloads/search-downloads/solr-6.6.0/server/solr/configsets/data_driven_schema_configs/conf
> for config gettingstarted to ZooKeeper at localhost:9983
> Creating new collection 'gettingstarted' using command:
> http://localhost:8983/solr/admin/collections?action=CREATE&name=gettingstarted&numShards=2&replicationFactor=2&maxShardsPerNode=2&collection.configName=gettingstarted{code}
> maxShardsPerNode get's set to 2 .
>
> Compare this to Solr 7.3
> {code:java}
> INFO - 2018-06-20 13:55:33.823;
> org.apache.solr.client.solrj.impl.ZkClientClusterStateProvider; Cluster at
> localhost:9983 ready
> Created collection 'gettingstarted' with 2 shard(s), 2 replica(s) with
> config-set 'gettingstarted'{code}
> So something changed and now we no longer set maxShardsPerNode and it
> defaults to -1 .
>
> -1 has special handing while creating a collection ( it means max int ) .
> This special handling is not there while Restoring a collection and hence
> this fails
> We should not set maxShardsPerNode to -1 in the first place
> Steps to reproduce:
> # ./bin/solr start -e cloud -noprompt : This creates a 2 node cluster and a
> gettingstarted collection which 2X2
> 2. Added 4 docs (id=1,2,3,4) with commit=true and openSearcher=true (default)
> 3. Call backup:
> http://localhost:8983/solr/admin/collections?action=BACKUP&name=gettingstarted_backup&collection=gettingstarted&location=/Users/varunthacker/solr-7.1.0
> 4. Call restore:
> [http://localhost:8983/solr/admin/collections?action=restore&name=gettingstarted_backup&collection=restore_gettingstarted&location=/Users/varunthacker/solr-7.1.0]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
