[
https://issues.apache.org/jira/browse/SOLR-12291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16458279#comment-16458279
]
Varun Thacker edited comment on SOLR-12291 at 4/30/18 2:46 AM:
---------------------------------------------------------------
Here's a simple way to reproduce this:
Start Solr with
{code:java}
bin/solr start -e gettingstarted -noprompt{code}
Call collection reload with async
{code:java}
http://localhost:8983/solr/admin/collections?action=reload&name=gettingstarted&async=1{code}
Call request status
{code:java}
http://localhost:8983/solr/admin/collections?action=requeststatus&requestid=1{code}
Here's the output ( it's a miracle our users are able to make some sense of
this response ;) )
{code:java}
{
"responseHeader":{
"status":0,
"QTime":5},
"success":{
"192.168.0.6:8983_solr":{
"responseHeader":{
"status":0,
"QTime":1}},
"192.168.0.6:8983_solr":{
"responseHeader":{
"status":0,
"QTime":0}},
"192.168.0.6:7574_solr":{
"responseHeader":{
"status":0,
"QTime":58}},
"192.168.0.6:7574_solr":{
"responseHeader":{
"status":0,
"QTime":59}}},
"167115373362154":{
"responseHeader":{
"status":0,
"QTime":0},
"STATUS":"completed",
"Response":"TaskId: 167115373362154 webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n4&async=167115373362154&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=58"},
"167115373632133":{
"responseHeader":{
"status":0,
"QTime":0},
"STATUS":"completed",
"Response":"TaskId: 167115373632133 webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n6&async=167115373632133&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=0"},
"status":{
"state":"completed",
"msg":"found [1] in completed tasks"}}{code}
Notice how we only track 2 task id's ..It should be 4 because the
gettingstarted example creates a 2X2 collection.
Also if you check the logs there were 4 tasks but the overseer didn't track 2
of them
{code:java}
~/solr-7.3.0$ grep "/admin/cores" example/cloud/node1/logs/solr.log | grep -i
reload
INFO - 2018-04-30 02:40:05.614; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard1_replica_n2&async=167115373148039&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=1
INFO - 2018-04-30 02:40:05.614; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n6&async=167115373632133&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=0
~/solr-7.3.0$ grep "/admin/cores" example/cloud/node2/logs/solr.log | grep -i
reload
INFO - 2018-04-30 02:40:05.672; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard1_replica_n1&async=167115372921360&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=59
INFO - 2018-04-30 02:40:05.672; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n4&async=167115373362154&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=58{code}
was (Author: varunthacker):
Here's a simple way to reproduce this:
Start Solr with
{code:java}
bin/solr start -e gettingstarted -noprompt{code}
Call collection reload with async
{code:java}
http://localhost:8983/solr/admin/collections?action=reload&name=gettingstarted&async=1{code}
Call request status
{code:java}
http://localhost:8983/solr/admin/collections?action=requeststatus&requestid=1{code}
Here's the output
{code:java}
{
"responseHeader":{
"status":0,
"QTime":5},
"success":{
"192.168.0.6:8983_solr":{
"responseHeader":{
"status":0,
"QTime":1}},
"192.168.0.6:8983_solr":{
"responseHeader":{
"status":0,
"QTime":0}},
"192.168.0.6:7574_solr":{
"responseHeader":{
"status":0,
"QTime":58}},
"192.168.0.6:7574_solr":{
"responseHeader":{
"status":0,
"QTime":59}}},
"167115373362154":{
"responseHeader":{
"status":0,
"QTime":0},
"STATUS":"completed",
"Response":"TaskId: 167115373362154 webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n4&async=167115373362154&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=58"},
"167115373632133":{
"responseHeader":{
"status":0,
"QTime":0},
"STATUS":"completed",
"Response":"TaskId: 167115373632133 webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n6&async=167115373632133&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=0"},
"status":{
"state":"completed",
"msg":"found [1] in completed tasks"}}{code}
Notice how we only track 2 task id's ..It should be 4 because the
gettingstarted example creates a 2X2 collection.
Also if you check the logs there were 4 tasks but the overseer didn't track 2
of them
{code:java}
~/solr-7.3.0$ grep "/admin/cores" example/cloud/node1/logs/solr.log | grep -i
reload
INFO - 2018-04-30 02:40:05.614; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard1_replica_n2&async=167115373148039&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=1
INFO - 2018-04-30 02:40:05.614; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n6&async=167115373632133&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=0
~/solr-7.3.0$ grep "/admin/cores" example/cloud/node2/logs/solr.log | grep -i
reload
INFO - 2018-04-30 02:40:05.672; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard1_replica_n1&async=167115372921360&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=59
INFO - 2018-04-30 02:40:05.672; [ ] org.apache.solr.servlet.HttpSolrCall;
[admin] webapp=null path=/admin/cores
params={core=gettingstarted_shard2_replica_n4&async=167115373362154&qt=/admin/cores&action=RELOAD&wt=javabin&version=2}
status=0 QTime=58{code}
> OverseerCollectionMessageHandler sliceCmd assumes only one replica exists on
> each node
> --------------------------------------------------------------------------------------
>
> Key: SOLR-12291
> URL: https://issues.apache.org/jira/browse/SOLR-12291
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: Backup/Restore, SolrCloud
> Reporter: Varun Thacker
> Priority: Major
>
> The OverseerCollectionMessageHandler sliceCmd assumes only one replica exists
> on one node
> When multiple replicas of a slice are on the same node we only track one
> replica's async request. This happens because the async requestMap's key is
> "node_name"
> I discovered this when [~alabax] shared some logs of a restore issue, where
> the second replica got added before the first replica had completed it's
> restorecore action.
> While looking at the logs I noticed that the overseer never called
> REQUESTSTATUS for the restorecore action , almost as if it had missed
> tracking that particular async request.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
