[
https://issues.apache.org/jira/browse/USERGRID-1052?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14971756#comment-14971756
]
Todd Nine commented on USERGRID-1052:
-------------------------------------
This was caused by several issues that have been address in this fix.
1) Empty shards need read-repaired and removed
2) When graph edges are missed during the delete event, they have to all be
iterated and skipped. This needs fixed in 2 areas.
2.1) The delete event needs to properly compact the node in the graph
2.2) The read needs to catch these issues and fire the repair event.
> After large deletions, graph seek time becomes slow
> ---------------------------------------------------
>
> Key: USERGRID-1052
> URL: https://issues.apache.org/jira/browse/USERGRID-1052
> Project: Usergrid
> Issue Type: Bug
> Reporter: Todd Nine
> Assignee: Todd Nine
> Labels: landmine
>
> After a large number of deletions, 100k +, seek time on graph becomes very
> slow when searching collections. Identify the issue, and implement a fix
> based off of sha a09485a3a5ac8b4217b294f4754ea8a70a7ec447
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
