[
https://issues.apache.org/jira/browse/HBASE-1342?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12704791#action_12704791
]
stack commented on HBASE-1342:
------------------------------
Currently, I'm thinking of writing a file into the Region subdirectory named
regioninfo on region creation. It would contain the serialized HRegionInfo
followed by a text version of HRegionInfo. Cat'ing the regioninfo file, you'd
see binary garbage followed by text representation. A tool should be able to
recreate an HRI just be deserializing from this file. The text part would be
so humans could see what the region was about.
I was originally thinking of writing this file every time the region was
deployed. This would up the hdfs load on startup. I was thinking we'd do this
because parts of the HRI can go stale and rewriting it on every deploy would
keep it pretty fresh. But I think as far as rebuilding the .META., figuring
what the table and column descriptors were just before the need for repair will
be figured elsewhere.
A tool that rebuilds the .META. will still not have enough to rebuild the
.META. using regioninfo. The rebuild tool will have to sort and make sense of
the list of regions it has figuring which are parent regions and which are
daughters, etc., but I think that this can be done up in the tool using common
sense checking if storefiles have references, etc.; in other words, there is
enough to go on rebuilding table and column descriptors. The regioninfo is for
region name, region id and for easy access to start/end key, critical info
needed rebuilding .META.
Was also going to change master logging so it emitted the HRI at INFO level so
could go there ultimately to resolve what happened when -- if needed.
> Add to filesystem info needed to rebuild .META.
> -----------------------------------------------
>
> Key: HBASE-1342
> URL: https://issues.apache.org/jira/browse/HBASE-1342
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Fix For: 0.20.0
>
>
> From Cosmin up on list:
> {code}
> Hi,
> Is there any way to rebuild the .META. Table with information in HDFS?
> Long story short, the namenode ran out of disk and crashed and we couldn't
> have it load the edits.new file at startup (the edits.new file seemed
> corrupted and it failed to read some Long value) so we removed the edits.new
> file. Apparently the edits file had 10 - 15 entries. Some edit data on
> -ROOT-, some edit data on .META., some data on a table compaction
> (comapction.dir) and some data on some hbase logs. Only last entry seem to
> be bogus, having almost a megabyte of trailing zeroes.
> After renaming edits.new HDFS started ok and HBase data is present. However
> when we started Hbase one table was missing entirely and one table was
> empty. The HBase master report shows Regions On FS: 94, but the report on
> region servers shows only 10 regions on all the region servers.
> Is there any way to cleanup hbase and rebuild the .META. table from HDFS?
> This would be ok even if it implies throwing up some data.
> If we would clean the edits.new file and restart HDFS or if we would
> manually apply the edits would that still make consistent changes to HBase
> data?
> Thanks,
> Cosmin
> {code}
> Lets log the info needed to rebuild your .META. or add things like
> HRegionInfo made creating a region as a file under each region directory.
> That don't seem hard and the info could be used rebuilding. What else would
> we need?
> Marking as blocker/critical for 0.20.0.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
