lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Shalin Shekhar Mangar (JIRA)" <>
Subject [jira] [Updated] (SOLR-6554) Speed up overseer operations for collections with stateFormat > 1
Date Sat, 29 Nov 2014 14:51:14 GMT


Shalin Shekhar Mangar updated SOLR-6554:
    Attachment: SOLR-6554.patch

Here's a patch which adds batching of operations for stateFormat=2 as along as the consecutive
messages operate on the same collection. It is not possible to preserve atomicity of writes
for different collections because they are written to multiple zk nodes.

I think this is ready. I'll commit this patch to trunk as a checkpoint. But I want to refactor
a bit more such that the batching can be encapsulated inside ZkStateWriter further simplifying
the overseer loop.

> Speed up overseer operations for collections with stateFormat > 1
> -----------------------------------------------------------------
>                 Key: SOLR-6554
>                 URL:
>             Project: Solr
>          Issue Type: Improvement
>          Components: SolrCloud
>    Affects Versions: 5.0, Trunk
>            Reporter: Shalin Shekhar Mangar
>         Attachments: SOLR-6554.patch, SOLR-6554.patch, SOLR-6554.patch, SOLR-6554.patch,
SOLR-6554.patch, SOLR-6554.patch, SOLR-6554.patch
> Right now (after SOLR-5473 was committed), a node watches a collection only if stateFormat=1
or if that node hosts at least one core belonging to that collection.
> This means that a node which is the overseer operates on all collections but watches
only a few. So any read goes directly to zookeeper which slows down overseer operations.
> Let's have the overseer node watch all collections always and never remove those watches
(except when the collection itself is deleted).

This message was sent by Atlassian JIRA

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message