Got it! So I will re-run a randomWrite (to populate the table) and a randomRead with 64 bytes cells. Should that show the difference between the 2?
2014-10-02 13:25 GMT-04:00 Vladimir Rodionov <[email protected]>: > Any read/scan from block cache test should show some improvement with > small KVs (1Kb is large). > > -Vladimir Rodionov > > On Thu, Oct 2, 2014 at 10:08 AM, Jean-Marc Spaggiari < > [email protected]> wrote: > > > Hi Lars, > > > > What is the exact use-case where I should see improvement? I can add it > in > > my test script and re-run that part on 0.94.23 to compare. > > > > So 1k KVs, but most on the reads? On the writes? On scanning? Skipping? > > > > Thanks, > > > > JM > > > > 2014-10-02 12:43 GMT-04:00 lars hofhansl <[email protected]>: > > > > > Thanks JM. > > > > > > > > > Looking forward to some perf numbers. 0.94.24 should have a bunch of > > > improvements. (Although PE will probably not show them as it operates > on > > 1k > > > KVs, and so perf is dominated by moving the KVs around.) > > > > > > -- Lars > > > > > > > > > > > > ----- Original Message ----- > > > From: Jean-Marc Spaggiari <[email protected]> > > > To: dev <[email protected]> > > > Cc: lars hofhansl <[email protected]> > > > Sent: Wednesday, October 1, 2014 6:35 PM > > > Subject: Re: The 3rd hbase 0.94.24 release candidate is available for > > > download > > > > > > Downloaded, uncompressed, signature checked => Passed > > > Checked files (.TXT) and doc => Broken linkes in README.txt and 2.3.1 > is > > > malformated. > > > Started standalone, checked logs and UI => Passed. > > > > > > Test suite JDK1.7 => Passed twice, failed once: > > > Failed tests: > > > > > > > > > testExceptionFromCoprocessorDuringPut(org.apache.hadoop.hbase.coprocessor.TestRegionServerCoprocessorExceptionWithAbort): > > > The put should have failed, as the coprocessor is buggy > > > > > > > > > > > > testMultiRowMutationMultiThreads(org.apache.hadoop.hbase.regionserver.TestAtomicOperation): > > > expected:<0> but was:<1> > > > > > > Test suite JDK1.6 => Passed twice in a row. > > > Create table, scan, put, get, disable, drop => Passed. > > > > > > Looks good to me too. +1 > > > > > > Perfs tests are running, I will report tomorrow or Friday. > > > > > > JM > > > > > > > > > > > > > > > 2014-10-01 19:41 GMT-04:00 Ted Yu <[email protected]>: > > > > > > > Verified signtaure. > > > > Ran test suite - TestSnapshotFromMaster#testSnapshotHFileArchiving > > failed > > > > but passed when run alone. > > > > Done some random shell ops. > > > > All looks good > > > > > > > > +1 > > > > > > > > On Wed, Oct 1, 2014 at 2:03 PM, lars hofhansl <[email protected]> > > wrote: > > > > > > > > > +1 from me > > > > > > > > > > - Installed the binary release. > > > > > - checked signature > > > > > - checked docs, etc > > > > > - inserted some data > > > > > - flushed, compacted, scanned, etc, all good > > > > > - nothing strange in the log* > > > > > > > > > > > > > > > * I did find a message logged where we apparently do not escape > > binary > > > > > data correctly. That's not a new problem, though. Logged > HBASE-12146 > > > for > > > > > that. > > > > > > > > > > > > > > > Please do not forget to vote for the release. Only two days left. > :) > > > > > > > > > > > > > > > -- Lars > > > > > > > > > > ________________________________ > > > > > From: lars hofhansl <[email protected]> > > > > > To: Dev <[email protected]> > > > > > Sent: Monday, September 29, 2014 6:03 PM > > > > > Subject: The 3rd hbase 0.94.24 release candidate is available for > > > > download > > > > > > > > > > > > > > > The 3rd (and hopefully last) 0.94.24 RC is available for download > at > > > > > http://people.apache.org/~larsh/hbase-0.94.24-rc2/ > > > > > Signed with my code signing key: C7CFE328 > > > > > > > > > > HBase 0.94.24 is a bug fix release with 13 fixes: > > > > > [HBASE-11923] - Potential race condition in > > > > > RecoverableZookeeper.checkZk() > > > > > [HBASE-11963] - Synchronize peer cluster replication connection > > > > > attempts > > > > > [HBASE-12023] - HRegion.applyFamilyMapToMemstore creates too > many > > > > > iterator objects. > > > > > [HBASE-12077] - FilterLists create many ArrayList$Itr objects > per > > > > row. > > > > > [HBASE-11405] - Multiple invocations of hbck in parallel > disables > > > > > balancer permanently > > > > > [HBASE-11957] - Backport to 0.94 HBASE-5974 Scanner retry > > behavior > > > > > with RPC timeout on next() seems incorrect > > > > > [HBASE-12019] - hbase-daemon.sh overwrite HBASE_ROOT_LOGGER and > > > > > HBASE_SECURITY_LOGGER variables > > > > > [HBASE-12020] - String formatting on each RPC Invoke > > > > > [HBASE-12022] - Payloads on Failure attempt to serialize the > > byte[] > > > > > into strings. > > > > > [HBASE-12114] - Meta table cache hashing may access the wrong > > table > > > > > [HBASE-12090] - Bytes: more Unsafe, more Faster > > > > > [HBASE-12103] - Backport HFileV1Detector to 0.94 > > > > > [HBASE-12113] - Backport to 0.94: HBASE-5525 Truncate and > > preserve > > > > > region boundaries option > > > > > > > > > > The list of changes is also available here: > > > > > > > > > > > > > > > https://issues.apache.org/jira/secure/ReleaseNote.jspa?projectId=12310753&version=12327569 > > > > > > > > > > Here's the test run for this RC: > > > > > https://builds.apache.org/job/HBase-0.94.24/43/ > > > > > > > > > > Please try out the RC, check out the doc, take it for a spin, etc, > > and > > > > > vote +1/-1 by EOD October 3rd on whether we should release this as > > > > 0.94.24. > > > > > > > > > > Thanks. > > > > > > > > > > -- Lars > > > > > > > > > > > > > > > > > >
