[
https://issues.apache.org/jira/browse/SOLR-7982?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Forest Soup updated SOLR-7982:
------------------------------
Description:
We have a 3 Zookeeper 5 solr server Solrcloud.
We created collection1 and collection2 with 80 shards respectively in the
cloud, replicateFactor is 2.
But after created, we found in a same collection, the coreNodeName has some
duplicate in core.properties in the core folder. For example:
[tanglin@solr64 home]$ ll collection1_shard13_replica2/core.properties
-rw-r--r-- 1 solr solr 173 Jul 29 11:52
collection1_shard13_replica2/core.properties
[tanglin@solr64 home]$ ll collection1_shard66_replica1/core.properties
-rw-r--r-- 1 solr solr 173 Jul 29 11:52
collection1_shard66_replica1/core.properties
[tanglin@solr64 home]$ cat collection1_shard66_replica1/core.properties
#Written by CorePropertiesLocator
#Wed Jul 29 11:52:54 UTC 2015
numShards=80
name=collection1_shard66_replica1
shard=shard66
collection=collection1
coreNodeName=core_node19
[tanglin@solr64 home]$ cat collection1_shard13_replica2/core.properties
#Written by CorePropertiesLocator
#Wed Jul 29 11:52:53 UTC 2015
numShards=80
name=collection1_shard13_replica2
shard=shard13
collection=collection1
coreNodeName=core_node19
[tanglin@solr64 home]$
The consequence of the issue is that the clusterstate.json in zookeeper is also
with wrong core_node#, and updating state of a core sometimes changed the state
of other core in other shard..
Snippet from clusterstate:
"shard13":{
"range":"a6660000-a998ffff",
"state":"active",
"replicas":{
"core_node33":{
"state":"active",
"base_url":"https://solr65.somesite.com:8443/solr",
"core":"collection1_shard13_replica1",
"node_name":"solr65.somesite.com:8443_solr"},
"core_node19":{
"state":"active",
"base_url":"https://solr64.somesite.com:8443/solr",
"core":"collection1_shard13_replica2",
"node_name":"solr64.somesite.com:8443_solr",
"leader":"true"}}},
...
"shard66":{
"range":"50000000-5332ffff",
"state":"active",
"replicas":{
"core_node105":{
"state":"active",
"base_url":"https://solr63.somesite.com:8443/solr",
"core":"collection1_shard66_replica2",
"node_name":"solr63.somesite.com:8443_solr",
"leader":"true"},
"core_node19":{
"state":"active",
"base_url":"https://solr64.somesite.com:8443/solr",
"core":"collection1_shard66_replica1",
"node_name":"solr64.somesite.com:8443_solr"}}},
was:
We have a 3 Zookeeper 5 solr server Solrcloud.
We created collection1 and collection2 with 80 shards respectively in the
cloud, replicateFactor is 2.
But after created, we found in a same collection, the coreNodeName has some
duplicate in core.properties in the core folder. For example:
[tanglin@solr64 home]$ ll collection1_shard13_replica2/core.properties
-rw-r--r-- 1 solr solr 173 Jul 29 11:52
collection1_shard13_replica2/core.properties
[tanglin@solr64 home]$ ll collection1_shard66_replica1/core.properties
-rw-r--r-- 1 solr solr 173 Jul 29 11:52
collection1_shard66_replica1/core.properties
[tanglin@solr64 home]$ cat collection1_shard66_replica1/core.properties
#Written by CorePropertiesLocator
#Wed Jul 29 11:52:54 UTC 2015
numShards=80
name=collection1_shard66_replica1
shard=shard66
collection=collection1
coreNodeName=core_node19
[tanglin@solr64 home]$ cat collection1_shard13_replica2/core.properties
#Written by CorePropertiesLocator
#Wed Jul 29 11:52:53 UTC 2015
numShards=80
name=collection1_shard13_replica2
shard=shard13
collection=collection1
coreNodeName=core_node19
[tanglin@solr64 home]$
The consequence of the issue is that the clusterstate.json in zookeeper is also
with wrong core_node#, and updating state of a core sometimes changed the state
of other core in other shard..
Snippet from clusterstate:
"shard13":{
"range":"a6660000-a998ffff",
"state":"active",
"replicas":{
"core_node33":{
"state":"active",
"base_url":"https://us1a3-solr65.a3.dal06.isc4sb.com:8443/solr",
"core":"collection1_shard13_replica1",
"node_name":"us1a3-solr65.a3.dal06.isc4sb.com:8443_solr"},
"core_node19":{
"state":"active",
"base_url":"https://us1a3-solr64.a3.dal06.isc4sb.com:8443/solr",
"core":"collection1_shard13_replica2",
"node_name":"us1a3-solr64.a3.dal06.isc4sb.com:8443_solr",
"leader":"true"}}},
...
"shard66":{
"range":"50000000-5332ffff",
"state":"active",
"replicas":{
"core_node105":{
"state":"active",
"base_url":"https://us1a3-solr63.a3.dal06.isc4sb.com:8443/solr",
"core":"collection1_shard66_replica2",
"node_name":"us1a3-solr63.a3.dal06.isc4sb.com:8443_solr",
"leader":"true"},
"core_node19":{
"state":"active",
"base_url":"https://us1a3-solr64.a3.dal06.isc4sb.com:8443/solr",
"core":"collection1_shard66_replica1",
"node_name":"us1a3-solr64.a3.dal06.isc4sb.com:8443_solr"}}},
> SolrCloud: collection creation: There are duplicate coreNodeName in
> core.properties in a same collection.
> ---------------------------------------------------------------------------------------------------------
>
> Key: SOLR-7982
> URL: https://issues.apache.org/jira/browse/SOLR-7982
> Project: Solr
> Issue Type: Bug
> Components: SolrCloud
> Affects Versions: 4.7
> Environment: linux redhat enterprise server 5.9 64bit
> Reporter: Forest Soup
>
> We have a 3 Zookeeper 5 solr server Solrcloud.
> We created collection1 and collection2 with 80 shards respectively in the
> cloud, replicateFactor is 2.
> But after created, we found in a same collection, the coreNodeName has some
> duplicate in core.properties in the core folder. For example:
> [tanglin@solr64 home]$ ll collection1_shard13_replica2/core.properties
> -rw-r--r-- 1 solr solr 173 Jul 29 11:52
> collection1_shard13_replica2/core.properties
> [tanglin@solr64 home]$ ll collection1_shard66_replica1/core.properties
> -rw-r--r-- 1 solr solr 173 Jul 29 11:52
> collection1_shard66_replica1/core.properties
> [tanglin@solr64 home]$ cat collection1_shard66_replica1/core.properties
> #Written by CorePropertiesLocator
> #Wed Jul 29 11:52:54 UTC 2015
> numShards=80
> name=collection1_shard66_replica1
> shard=shard66
> collection=collection1
> coreNodeName=core_node19
> [tanglin@solr64 home]$ cat collection1_shard13_replica2/core.properties
> #Written by CorePropertiesLocator
> #Wed Jul 29 11:52:53 UTC 2015
> numShards=80
> name=collection1_shard13_replica2
> shard=shard13
> collection=collection1
> coreNodeName=core_node19
> [tanglin@solr64 home]$
> The consequence of the issue is that the clusterstate.json in zookeeper is
> also with wrong core_node#, and updating state of a core sometimes changed
> the state of other core in other shard..
> Snippet from clusterstate:
> "shard13":{
> "range":"a6660000-a998ffff",
> "state":"active",
> "replicas":{
> "core_node33":{
> "state":"active",
> "base_url":"https://solr65.somesite.com:8443/solr",
> "core":"collection1_shard13_replica1",
> "node_name":"solr65.somesite.com:8443_solr"},
> "core_node19":{
> "state":"active",
> "base_url":"https://solr64.somesite.com:8443/solr",
> "core":"collection1_shard13_replica2",
> "node_name":"solr64.somesite.com:8443_solr",
> "leader":"true"}}},
> ...
> "shard66":{
> "range":"50000000-5332ffff",
> "state":"active",
> "replicas":{
> "core_node105":{
> "state":"active",
> "base_url":"https://solr63.somesite.com:8443/solr",
> "core":"collection1_shard66_replica2",
> "node_name":"solr63.somesite.com:8443_solr",
> "leader":"true"},
> "core_node19":{
> "state":"active",
> "base_url":"https://solr64.somesite.com:8443/solr",
> "core":"collection1_shard66_replica1",
> "node_name":"solr64.somesite.com:8443_solr"}}},
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]