Thanks Eric,

Yeah, running compact cleared out the entries.  I thought accumulo would do
that on it's own if I left it alone long enough.  Minor compaction was
running when I closed my laptop, but zookeeper got in a bad state
overnight.  Maybe it would have cleaned up,  I guess I never noticed it
before.

I just tested on install on RHEL6 and get the same behavior.

So from my testing, I don't see any problems on Mac OSX 1.6 other than my
understanding.

Mike

On Thu, Nov 10, 2011 at 6:46 AM, Eric Newton <[email protected]> wrote:

> Well, a delete is really "insert a delete marker" so, yes, the count can go
> up.  You can clear everything up with another compaction
>
> shell> compact junkTable
> shell>
>
> The functional tests passed on 1.3.5rc6... and they check a lot more
> input/output/restart conditions.
>
> Unfortunately, I didn't see the screenshot attached.
>
> -Eric
>
> On Thu, Nov 10, 2011 at 6:05 AM, Michael Wall <[email protected]> wrote:
>
> > Same thing this morning, still 15 Million entries in the test_ingest
> table
> > and 2 entries in the junk table.
> >
> > I went back to an earlier version and ran the same test and get the same
> > results.  Is my assumption that the # entries shown on localhost:50095
> > decrement on a delete incorrect?  Is this the functionality on other
> OS's?
> >
> > Mike
> >
> > On Wed, Nov 9, 2011 at 9:29 PM, Michael Wall <[email protected]> wrote:
> >
> > > +1 md5
> > > +1 mvn clean package && mvn assembly:single
> > >
> > > reran test/system/test1 again.  Missed the part in the README about
> > > creating that table, so I did that this time.  Much better results, the
> > > args for the CreateTable needed the username and password added.  Here
> is
> > > what I ran:
> > > /opt/accumulo-test/accumulo/bin/accumulo
> > > 'org.apache.accumulo.server.test.TestIngest$CreateTable' 0 5000000 100
> > root
> > > secret
> > >
> > > I realized that the username and password are hardcoded for the tests,
> so
> > > it would not work if you had something different than the defaults.
> > >
> > > After ingest_test.sh, verify_test.sh, ingest_test_2.sh and
> > > verify_test_2.sh I could scan the test_ingest table just fine.  The
> > > get_auths command returns G1,GROUP2,L1,L2
> > >
> > > However, after running test_ingest_3.sh, scan still takes several
> seconds
> > > and returns no results.   Here is the last part of the output from
> > > ingest_test_3.sh
> > > --------------------------------------
> > > 09 21:04:10,387 [impl.TabletServerBatchWriter] TRACE:
> > > 09 21:04:10,387 [impl.TabletServerBatchWriter] TRACE: TABLET SERVER
> BATCH
> > > WRITER STATISTICS
> > > 09 21:04:10,387 [impl.TabletServerBatchWriter] TRACE: Added
> > >  :  1,000,000 mutations
> > > 09 21:04:10,387 [impl.TabletServerBatchWriter] TRACE: Sent
> > > :  1,000,000 mutations
> > > 09 21:04:10,387 [impl.TabletServerBatchWriter] TRACE: Resent percentage
> > > :       0.00%
> > > 09 21:04:10,387 [impl.TabletServerBatchWriter] TRACE: Overall time
> > > :      82.90 secs
> > > 09 21:04:10,387 [impl.TabletServerBatchWriter] TRACE: Overall send rate
> > >  :  12,062.44 mutations/sec
> > > 09 21:04:10,388 [impl.TabletServerBatchWriter] TRACE: Send efficiency
> > >  :      97.30%
> > > 09 21:04:10,388 [impl.TabletServerBatchWriter] TRACE:
> > > 09 21:04:10,388 [impl.TabletServerBatchWriter] TRACE: BACKGROUND WRITER
> > > PROCESS STATISTICS
> > > 09 21:04:10,388 [impl.TabletServerBatchWriter] TRACE: Total send time
> > >  :      80.66 secs  97.30%
> > > 09 21:04:10,388 [impl.TabletServerBatchWriter] TRACE: Average send rate
> > >  :  12,397.72 mutations/sec
> > > 09 21:04:10,388 [impl.TabletServerBatchWriter] TRACE: Total bin time
> > > :       1.81 secs   2.18%
> > > 09 21:04:10,388 [impl.TabletServerBatchWriter] TRACE: Average bin rate
> > > : 553,097.35 mutations/sec
> > > 09 21:04:10,389 [impl.TabletServerBatchWriter] TRACE: tservers per
> batch
> > > :     1.00 avg       1 min      1 max
> > > 09 21:04:10,389 [impl.TabletServerBatchWriter] TRACE: tablets per batch
> > >  :     1.57 avg       1 min      2 max
> > > 09 21:04:10,389 [impl.TabletServerBatchWriter] TRACE:
> > > 09 21:04:10,389 [impl.TabletServerBatchWriter] TRACE: SYSTEM STATISTICS
> > > 09 21:04:10,389 [impl.TabletServerBatchWriter] TRACE: JVM GC Time
> > >  :       1.11 secs
> > > 09 21:04:10,389 [impl.TabletServerBatchWriter] TRACE: JVM Compile Time
> > > :      12.10 secs
> > > 09 21:04:10,390 [impl.TabletServerBatchWriter] TRACE: System load
> average
> > > : initial=  0.68 final=  3.85
> > >    1,000,000 records written |   12,060 records/sec |   92,000,000
> bytes
> > > written | 1,109,596 bytes/sec | 82.913 secs
> > > --------------------------------------------------------------
> > >
> > > Also attached is a screenshot of the monitor page showing the only
> tablet
> > > server I have.  It is still compacting, but I would expect the entries
> to
> > > be deleted by now.  Been compacting for 15 minutes.
> > >
> > > Everything else seemed fine.  I created a junk table in inserted one
> > > record from the shell and then deleted it.  Works fine, but the monitor
> > > shows 2 entries for that.  Maybe this is the normal behavior.  I can't
> > > troubleshoot any further tonight, but will let it run to see if it
> > > eventually cleans up.  Will check in the morning.
> > >
> > > Thanks
> > >
> > > Mike
> > >
> > > On Wed, Nov 9, 2011 at 9:29 AM, Keith Turner <[email protected]> wrote:
> > >
> > >> On Tue, Nov 8, 2011 at 8:33 AM, Eric Newton <[email protected]>
> > >> wrote:
> > >> > I propose that we accept
> > >> http://people.apache.org/~ecn/accumulo-1.3.5rc6/
> > >> > accumulo-1.3.5-incubating-rc6-dist.tar.gz
> > >> > as the official Accumulo 1.3.5 release.  In previous release
> > candidates
> > >> we
> > >> > found
> > >> >  * some bad links in the documentation
> > >> >  * fonts missing on the build machine resulting in bad characters in
> > the
> > >> > user manual
> > >> >  * there were build artifacts in the src directory
> > >> > See ACCUMULO-125.
> > >> >
> > >> > I generated this tarball by checking out
> > >> > https://svn.apache.org/repos/asf/incubator/accumulo/tags/1.3.5rc6
> > >> > and running "./src/assembly/build.sh"
> > >> >
> > >> > The GPG signature can be found at
> > >> > http://people.apache.org/~ecn/accumulo-1.3.5rc6/
> > >> > accumulo-1.3.5-incubating-rc6-dist.tar.gz.asc
> > >> >
> > >> > It has an MD5 sum of 9b7fff34869d28d2fc7bdc4da6d35a91.
> > >> >
> > >> > Please download, verify sig/sum, and install/test.
> > >> >
> > >> > This vote closes in 72 hours.
> > >> >
> > >>
> > >> We may want to look into running a link checker on the documentation
> > >> and monitor page as part of the release process.  Does anyone have any
> > >> suggestions on link checkers?
> > >>
> > >
> > >
> >
>

Reply via email to