hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jean-Daniel Cryans (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HBASE-533) Region Historian
Date Thu, 29 May 2008 13:28:45 GMT

    [ https://issues.apache.org/jira/browse/HBASE-533?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12600767#action_12600767
] 

Jean-Daniel Cryans commented on HBASE-533:
------------------------------------------

bq. I also wonder about listing parent events in the daughters. A daughter from a good family
who can trace her roots all the ways back to the pilgrims will have a long line in the table
above. I think listing just the immediate parent is sufficient, at least for the first version.

Or we can put the depth of history to keep as a parameter? Static or dynamic? For the moment,
I think it could just be a new conf in hbase-defaults.xml with value at 1.

bq. What about listing the event source? Most edits will be by the master but the splits will
be done by the splitting regionserver.

If it's constant eg splits always on regionserver and creation always by master, I don't think
we need it. I see a case where this would be a problem. For example, a user that never saw
the code and for some reason uses this information, he would need to know that. But since
this feature is for devs purpose and we know how it works, it shouldn't be a problem. Your
opinion?

bq. Regards double creations, I'd guess its not two creations - just two invocations of the
HRegion constructor. Opening and creating both call new HRegion. You might move our call to
addRegionCreation out of the HRegion constructor and instead put it into HRegion.createHRegion
(if this is appropriately called in all places a region is initially created).

I'll debug that.

bq. Dynamic would be nice - like log level? Tying it to loglevel would make your flag easy
to implement: e.g. send compaction/flush notice when at DEBUG. Bad thing about tying it to
loglevel doesn't have sufficient granularity. There's effectively INFO and DEBUG. What you
think?

Good idea, for the first version I'll do it like that. We'll see if we need more in the future.


Good stuff.

> Region Historian
> ----------------
>
>                 Key: HBASE-533
>                 URL: https://issues.apache.org/jira/browse/HBASE-533
>             Project: Hadoop HBase
>          Issue Type: Wish
>            Reporter: Bryan Duxbury
>            Assignee: Jean-Daniel Cryans
>            Priority: Minor
>         Attachments: hbase-533-v1-fix.patch, hbase-533-v1.patch
>
>
> Whenever we try to debug region splitting, assignment, compaction, etc. issues, we always
end up having to look in 1-20 different log files for cryptic names of regions and try to
piece together the chain of events ourselves. This is a challenging at best effort most of
the time.
> What would be very useful would be a new utility I've nicknamed the Region Historian.
You give it the text name of a region, and it will track down the log messages relevant to
it in the master and regionserver logs. Then, it will interleave the messages in such a way
that the timestamps correctly list the order of events. The result is a log summary that accurately
describes what happened to a region during it's lifetime, making it much easier to try and
figure out where something went wrong.
> Other things it could do would be replace cryptic log messages with simple events like
"the region was split into a and b", "the region was assigned to server x", and trace the
lineage of a region backwards to its parent before it came into existence.
> I'm sure there are other things we would think up that would be useful as well.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message