[jira] [Commented] (ACCUMULO-1770) out of memory error on very long running tablet server

["Eric Newton (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ACCUMULO-1770?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13792697#comment-13792697 ] 

Eric Newton commented on ACCUMULO-1770:
---------------------------------------

I ran an experiment.  I ingested random length values with a maximum size of N and minimum size N/2.  At the end of the run, I deleted the table and collected the tserver resident memory usage.  Each run ingested about 3.8G of value data.

||N||G bytes||
|256|	4.52|
|512|	4.50|
|1024|	4.50|
|2048|	4.48|
|4096|	4.50|
|8192|	4.51|
|16384|	4.52|
|32768|	4.52|
|65536|	4.54|
|131072|	4.54|
|262144|	4.54|
|524288|	4.56|
|1048576|	4.55|
|2097152|	4.55|
|4194304|	4.57|
|8388608|	4.60|
|16777216|	4.54|
|33554432|	4.39|

I'm not sure what to make of this.


> out of memory error on very long running tablet server
> ------------------------------------------------------
>
>                 Key: ACCUMULO-1770
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-1770
>             Project: Accumulo
>          Issue Type: Bug
>          Components: tserver
>            Reporter: Eric Newton
>            Assignee: Eric Newton
>         Attachments: memory-usage.png
>
>
> On a large cluster it was noticed that a few of the tablet servers had been pushed into swap.  This didn't effect the performance of the server until it ran out of memory, and the process was killed.  The gc reports in the debug log showed the system had plenty of heap space for the JVM.  The number of threads in the server were not excessive (dozens).  This cluster ingests some large values (megabytes).  The tablet server had been up for a month prior to running out of memory.  MALLOC_ARENA_MAX had already been set to 1.
> * Investigate the effect of fragmentation on memory usage for large value inserts.



--
This message was sent by Atlassian JIRA
(v6.1#6144)