[
https://issues.apache.org/jira/browse/ACCUMULO-4112?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15187237#comment-15187237
]
ASF GitHub Bot commented on ACCUMULO-4112:
------------------------------------------
Github user keith-turner commented on a diff in the pull request:
https://github.com/apache/accumulo/pull/79#discussion_r55533890
--- Diff:
server/tserver/src/main/java/org/apache/accumulo/tserver/TabletServer.java ---
@@ -2966,13 +2966,26 @@ public Void run() {
}
}
+ private Durability getMincEventDurability(KeyExtent extent) {
+ TableConfiguration conf;
+ if (extent.isMeta() || extent.isRootTablet()) {
--- End diff --
> Is there some other setting for user tables that's more appropriate?
This is the best setting I know of... Since the walog minc entries are used
during recovery, its nice to have them at the same sync level as the tablets
metadata.
> so the || extent.isRootTablet() is redundant
Nice catch, I wrongly assumed isMeta() was a more narrow check. I used to
know it did a wider check, but had forgotten.
> I think refactoring KeyExtent#isMeta is more appropriate
Changing its name would need to be a separate PR, because it would be very
noisy drowning out the changes for 4112. I agree another name that implies
more than `accumulo.metadata` would be good, however I am not sure what that
name would be.
> Poor performance due to MinC start/stop updates are always hsync'd
> ------------------------------------------------------------------
>
> Key: ACCUMULO-4112
> URL: https://issues.apache.org/jira/browse/ACCUMULO-4112
> Project: Accumulo
> Issue Type: Bug
> Components: tserver
> Environment: Fluo testing on a 20-node cluster
> Reporter: Eric Newton
> Assignee: Keith Turner
> Priority: Critical
> Fix For: 1.7.2, 1.8.0
>
> Attachments: HSyncOverheadExperiment.png, MinCFlushPerfTest.java,
> Sync-Flush-Log-Performance.png
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> [~kturner] writes:
> {quote}
> I was running a Fluo test with 1.8.0-SNAP on my workstation. My Fluo table
> had a ton of tablets. I was seeing terrible performance. I started
> looking at the tserver and noticed it was always calling hsync. I tracked
> down the problem to the fact that when minc start and stop events are written
> to the log they are always written w/ sync level. My poor little tserver
> was constantly minor compacting (probably had around 600 tablets that were
> all being written to).
> I changed the test config to create like 15 tablets and the performance was
> much better. All cores were 100% utilized, which was not the case when hsync
> was always called.
> {quote}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
