[ https://issues.apache.org/jira/browse/ACCUMULO-1085?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13600056#comment-13600056
]
Keith Turner commented on ACCUMULO-1085:
----------------------------------------
bq. Can we flush tablets that already have a lot buffered (i.e. those that are not being recovered)?
The problem I am thinking of is the following.
# Tserver A is hosting Tablet 1, Tablet 1 is using 80% of memory.
# Tserver B is hosting Tablet 2, Tablet 2 is using 80% of memory.
# Tserver A and Tserver B crash.
# Tablet server C is told to load Tablet 1 and Tablet 2.
# Tablet server C flushes all its tablets
# Tablet server C start loading Tablet 1 and Tablet 2 concurrently and crashes because it
runs out of memory.
We should probably figure out a way to handle this properly. But I am thinking we should
not do it in 1.5, unless its a very simple change that can be posted for review very quickly.
This is an improvement, and seems like it will only further delay 1.5.
> make the number of threads for assignment configurable
> ------------------------------------------------------
>
> Key: ACCUMULO-1085
> URL: https://issues.apache.org/jira/browse/ACCUMULO-1085
> Project: Accumulo
> Issue Type: Improvement
> Components: tserver
> Reporter: Eric Newton
> Assignee: Keith Turner
> Fix For: 1.6.0
>
> Attachments: ACCUMULO-1085.patch
>
>
> Log recoveries from sorted logs take considerable time: run them in parallel with more
assignment threads.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
|