lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF subversion and git services (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SOLR-11064) Collection APIs should provide disk space hint to Policy when possible
Date Tue, 16 Jan 2018 16:09:00 GMT

    [ https://issues.apache.org/jira/browse/SOLR-11064?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16327311#comment-16327311
] 

ASF subversion and git services commented on SOLR-11064:
--------------------------------------------------------

Commit ad829c3ea65a0f16884f7e1b8ad8a59c1c9466e8 in lucene-solr's branch refs/heads/branch_7x
from [~noble.paul]
[ https://git-wip-us.apache.org/repos/asf?p=lucene-solr.git;h=ad829c3 ]

SOLR-11064: Collection APIs should use the disk space hint when using policy framework


> Collection APIs should provide disk space hint to Policy when possible
> ----------------------------------------------------------------------
>
>                 Key: SOLR-11064
>                 URL: https://issues.apache.org/jira/browse/SOLR-11064
>             Project: Solr
>          Issue Type: Sub-task
>      Security Level: Public(Default Security Level. Issues are Public) 
>          Components: AutoScaling, SolrCloud
>            Reporter: Shalin Shekhar Mangar
>            Priority: Major
>             Fix For: master (8.0), 7.3
>
>
> SOLR-11063 added disk space hint to Policy. The ADDREPLICA, MOVEREPLICA, SPLITSHARD and
REPLACENODE APIs can provide this hint.
> # ADDREPLICA can query the size of the leader's data direcoty for the same shard and
pass it to policy
> # MOVEREPLICA again knows the size of the replica being moved
> # SPLITSHARD knows the size of the split indexes before it calls addreplica to create
replicas of the sub-shards
> # REPLACENODE can compute the sum of disk usage of all the replicas' data directories
> We should probably include both index and tlog size in the hint to ensure that we have
ample space on target nodes.
> We should also take care of HDFS because for movereplica and replacenode, there are no
disk space requirements because the original replica's HDFS data and ulog directories are
used by the new replicas.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message