[
https://issues.apache.org/jira/browse/HBASE-514?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12578509#action_12578509
]
Bryan Duxbury commented on HBASE-514:
-------------------------------------
I've been able to pretty easily fix the problem in the Memcache. However, I
think I've hit something of an algorithmic snag when it comes to sorting out
the closest row in the store files.
The problem lies in the fact that there's no guarantee newer store files are
written with data that comes after data in older store files. For instance, a
very recent store file could have data written from 10 minutes ago, but a store
file written 10 minutes ago could have data timestamped an hour in the future.
This means that you don't just need to check every store file for what it
thinks is the closest matching row. You actually have to ask every store file
what it thinks and then check it against *every other store file* to make sure
it isn't deleted! Basically, figuring out the closest row with the current
strategy is n^2 where n is the number of store files.
Maybe there's a better strategy to investigate. Perhaps if we ask every store
file what it thinks the best candidate is, then we sort the results by row, and
then scan all the store files from the first candidate to the last candidate,
noting any deletes as we go, and then ultimately choosing the maximum row
remaining in that list. Would that actually even help?
> table 'does not exist' when it does
> -----------------------------------
>
> Key: HBASE-514
> URL: https://issues.apache.org/jira/browse/HBASE-514
> Project: Hadoop HBase
> Issue Type: Bug
> Components: client
> Affects Versions: 0.16.0
> Reporter: stack
> Attachments: region-v2.patch
>
>
> This one I've seen a few times. In hql, I do show tables and it shows my
> table. I then try to do a select against the table and hql reports table
> does not exist. Digging, whats happening is that the getClosest facility is
> failing to find the first table region in the .META. table. I hacked up a
> region reading tool -- attached (for 0.1 branch) -- and tried it against but
> a copy and the actual instance of the region and it could do the getClosest
> fine. I'm pretty sure I restarted the HRS and when it came up again, the
> master had given it again the .META. and again was failing to find the first
> region in the table (Looked around in server logs and it seemed 'healthy').
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
