[jira] [Commented] (SOLR-12610) Inject failures during synchronous update requests during shard splits
[ https://issues.apache.org/jira/browse/SOLR-12610?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16574675#comment-16574675 ] Shalin Shekhar Mangar commented on SOLR-12610: -- I opened SOLR-12649 for the retry bug in CloudSolrClient > Inject failures during synchronous update requests during shard splits > -- > > Key: SOLR-12610 > URL: https://issues.apache.org/jira/browse/SOLR-12610 > Project: Solr > Issue Type: Test > Security Level: Public(Default Security Level. Issues are Public) > Components: SolrCloud >Reporter: Shalin Shekhar Mangar >Assignee: Shalin Shekhar Mangar >Priority: Major > Fix For: master (8.0), 7.5 > > Attachments: SOLR-12610-test-cloudclient-retry.patch, SOLR-12610.patch > > > In SOLR-12607, I found a bug where the StdNode's shard was not set correctly > causing exceptions during updates forwarded to sub-shard leaders to not be > sent back to the clients. This can cause data loss during split. A fix was > committed as part of SOLR-12607 but we need to expand coverage to this > situation. I'll add failure injection during the synchronous update step to > simulate this condition. This will be randomized for each shard split test > method. -- 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
[jira] [Commented] (SOLR-12610) Inject failures during synchronous update requests during shard splits
[ https://issues.apache.org/jira/browse/SOLR-12610?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16566452#comment-16566452 ] Shalin Shekhar Mangar commented on SOLR-12610: -- The new {{SOLR-12610-test-cloudclient-retry.patch}} has a test which tests retry behavior of CloudSolrClient. The HTTP 500 error from the server is not a communication problem and the cluster state is also unchanged and yet the CloudSolrClient attempts retries. > Inject failures during synchronous update requests during shard splits > -- > > Key: SOLR-12610 > URL: https://issues.apache.org/jira/browse/SOLR-12610 > Project: Solr > Issue Type: Test > Security Level: Public(Default Security Level. Issues are Public) > Components: SolrCloud >Reporter: Shalin Shekhar Mangar >Assignee: Shalin Shekhar Mangar >Priority: Major > Fix For: master (8.0), 7.5 > > Attachments: SOLR-12610-test-cloudclient-retry.patch, SOLR-12610.patch > > > In SOLR-12607, I found a bug where the StdNode's shard was not set correctly > causing exceptions during updates forwarded to sub-shard leaders to not be > sent back to the clients. This can cause data loss during split. A fix was > committed as part of SOLR-12607 but we need to expand coverage to this > situation. I'll add failure injection during the synchronous update step to > simulate this condition. This will be randomized for each shard split test > method. -- 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
[jira] [Commented] (SOLR-12610) Inject failures during synchronous update requests during shard splits
[ https://issues.apache.org/jira/browse/SOLR-12610?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16565465#comment-16565465 ] Erick Erickson commented on SOLR-12610: --- [~shalinmangar] Let me know if you'd like some beasting done on this either before or after you push it. > Inject failures during synchronous update requests during shard splits > -- > > Key: SOLR-12610 > URL: https://issues.apache.org/jira/browse/SOLR-12610 > Project: Solr > Issue Type: Test > Security Level: Public(Default Security Level. Issues are Public) > Components: SolrCloud >Reporter: Shalin Shekhar Mangar >Assignee: Shalin Shekhar Mangar >Priority: Major > Fix For: master (8.0), 7.5 > > Attachments: SOLR-12610.patch > > > In SOLR-12607, I found a bug where the StdNode's shard was not set correctly > causing exceptions during updates forwarded to sub-shard leaders to not be > sent back to the clients. This can cause data loss during split. A fix was > committed as part of SOLR-12607 but we need to expand coverage to this > situation. I'll add failure injection during the synchronous update step to > simulate this condition. This will be randomized for each shard split test > method. -- 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
[jira] [Commented] (SOLR-12610) Inject failures during synchronous update requests during shard splits
[ https://issues.apache.org/jira/browse/SOLR-12610?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16565168#comment-16565168 ] Shalin Shekhar Mangar commented on SOLR-12610: -- There are some failures with this randomization enabled. I see requests being retried by CloudSolrClient when there is no communication exception and no stale state. > Inject failures during synchronous update requests during shard splits > -- > > Key: SOLR-12610 > URL: https://issues.apache.org/jira/browse/SOLR-12610 > Project: Solr > Issue Type: Test > Security Level: Public(Default Security Level. Issues are Public) > Components: SolrCloud >Reporter: Shalin Shekhar Mangar >Assignee: Shalin Shekhar Mangar >Priority: Major > Fix For: master (8.0), 7.5 > > Attachments: SOLR-12610.patch > > > In SOLR-12607, I found a bug where the StdNode's shard was not set correctly > causing exceptions during updates forwarded to sub-shard leaders to not be > sent back to the clients. This can cause data loss during split. A fix was > committed as part of SOLR-12607 but we need to expand coverage to this > situation. I'll add failure injection during the synchronous update step to > simulate this condition. This will be randomized for each shard split test > method. -- 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