[
https://issues.apache.org/jira/browse/HBASE-2729?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12881117#action_12881117
]
HBase Review Board commented on HBASE-2729:
-------------------------------------------
Message from: "Todd Lipcon" <[email protected]>
bq. On 2010-06-22 00:14:05, stack wrote:
bq. > +1 (if it passes all tests). Nit-picks below.
Cool, I'll run it through my Hudson overnight, plus running a cluster test on
my 5-node test cluster now. Will commit tomorrow midday with changes below
addressed assuming testing goes OK.
bq. On 2010-06-22 00:14:05, stack wrote:
bq. > src/main/java/org/apache/hadoop/hbase/regionserver/Store.java, line 333
bq. > <http://review.hbase.org/r/216/diff/1/?file=1529#file1529line333>
bq. >
bq. > ? There was notion of a '_tmp' already?
bq. >
bq. > I'd say name it '.tmp'... since a '.' prefix seems to be our
convention given logs dir at top-level has a '.' prefix.
yea, the bulk load stuff uses a tmp dir if you ask the regionserver to load a
file which is stored on a different filesystem than the region itself.
I'll switch to .tmp though as you suggest (and move it into HConstants to
replace the now-unused compactions.dir or what have you)
bq. On 2010-06-22 00:14:05, stack wrote:
bq. > src/main/java/org/apache/hadoop/hbase/regionserver/Store.java, line 464
bq. > <http://review.hbase.org/r/216/diff/1/?file=1529#file1529line464>
bq. >
bq. > Why not keep old name and just move dirs? Why create a new unique
name?
I was worried that what's unique in the old dir might not be unique in the new
bq. On 2010-06-22 00:14:05, stack wrote:
bq. > src/test/java/org/apache/hadoop/hbase/regionserver/TestStore.java, line
411
bq. > <http://review.hbase.org/r/216/diff/1/?file=1531#file1531line411>
bq. >
bq. > Nice test.
thx! at some point I'd like to combine this with the one in TestFSErrorsExposed
and make a little mini-framework out of it... something for next month perhaps.
- Todd
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://review.hbase.org/r/216/#review261
-----------------------------------------------------------
> flushCache should write to a tmp directory and then move into the store
> directory
> ---------------------------------------------------------------------------------
>
> Key: HBASE-2729
> URL: https://issues.apache.org/jira/browse/HBASE-2729
> Project: HBase
> Issue Type: Bug
> Components: regionserver
> Affects Versions: 0.21.0
> Reporter: Todd Lipcon
> Assignee: stack
> Priority: Blocker
>
> Currently it appears that internalFlushCache writes directly to the target
> spot of the flushed data. The finally() block appends the metadata and closes
> the file as if nothing bad went wrong in case of an exception. This is really
> bad, since it means that an IOE in the middle of flushing cache could easily
> write a valid looking file with only half the data, which would then prevent
> us from recovering those edits during log replay.
> Instead, it should flush to a tmp location and move it into the region dir
> only after it's successfully written.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
