Hi J-D, I ran them twice.
The 2nd time 0.94.2 was 1.5% faster than 0.94.5. But as you are saying, I need to run that more to be able to get any conclusion. Now, this morning I have setup a dedicated computer to run those tests, have installed HBase 0.94.0 to 0.94.5 (6 versions) and scripted 6 tests in a batch file. Those tests will run for each version, and I will do that in loop something like 3 or 4 times. There is absolutly nothing else running on this computer. It's fully dedicated. Slow, but dedicated ;) I will post the results when it will be done. I'm not expecting this to be done before Friday because of the time it's taking. JM 2013/2/5, Jean-Daniel Cryans <jdcry...@apache.org>: > Are you running those tests more than once to eliminate variation? > > On Tue, Feb 5, 2013 at 4:14 AM, Jean-Marc Spaggiari > <jean-m...@spaggiari.org> wrote: >> Hi Lars, >> >> I saw that on the RandomSeekScanTest between 0.94.4 and 0.94.5. 2% is >> not big. This might have been because some other process start while I >> was running the tests. >> >> I have just the test. I will re-run all to compare. >> I have already put the results of the previous tests in the spreadsheet. >> >> Basically, >> 0.94.2 => 232 lines / seconds >> 0.94.4 => 231 lines / seconds >> 0.94.5 => 227 lines / seconds. >> >> 0.94.3 gave 228. >> >> I will update the spreadsheet when tests are done. >> >> JM >> >> 2013/2/4, lars hofhansl <la...@apache.org>: >>> Thanks JM, >>> >>> on what workload do you see the 2% slowdown? >>> >>> >>> -- Lars >>> >>> >>> >>> ________________________________ >>> From: Jean-Marc Spaggiari <jean-m...@spaggiari.org> >>> To: dev@hbase.apache.org; lars hofhansl <la...@apache.org> >>> Sent: Monday, February 4, 2013 10:35 AM >>> Subject: Re: ANN: The 1st hbase 0.94.5 release candidate is available >>> for >>> download >>> >>> Hi Lars, >>> >>> I gave it a big try and so far I don't see any issue. >>> >>> It's a bit slower than 0.94.4 (2%) but it might be related to my >>> tests. I will continue the performances tests and try YCSB on it if >>> time permit... >>> >>> I have updated the spreadsheet. >>> >>> So far, +1. >>> >>> JM >>> >>> 2013/2/2, lars hofhansl <la...@apache.org>: >>>> The 1st 0.94.5 RC is available for download at >>>> http://people.apache.org/~larsh/hbase-0.94.5-rc0/ >>>> >>>> Signed with my code signing key: C7CFE328 >>>> >>>> 0.94.5 is a bug fix release. This RC has 66 issues resolved against it, >>>> contributed by 26 individuals. >>>> These are mostly bug fixes. Also fixed is a long standing bug in >>>> replication >>>> along with lots of random test failures. >>>> >>>> The full list of changes is available here: >>>> https://issues.apache.org/jira/browse/HBASE/fixforversion/12323874 >>>> >>>> Please try out the RC, check out the doc, take it for a spin, etc, and >>>> vote >>>> +1/-1 by February 11th on whether we should release this as 0.94.5. >>>> >>>> The release testing spreadsheet is here: >>>> https://docs.google.com/spreadsheet/ccc?key=0AvLqcVIqNtlTdGNTUHVVY1lVZnpPT2Z5b0l2c3RoSHc#gid=0 >>>> (if you test this release, please add your test to this spreadsheet, so >>>> that >>>> we can gauge the coverage) >>>> >>>> HBase continues to be actively maintained; changes are coming in at a >>>> rate >>>> of about two commits per day. >>>> There is no slow down in this RC with 66 commits over the course of >>>> about >>>> a >>>> month. >>>> >>>> Thanks. >>>> >>>> -- Lars >
