Thanks Sandhya, very useful! BTW, if I want to change a test to take into account valid changes that translate into false failure, should I modify the comparison file, or the file that logs known legitimate differences? Thanks again for the help :-), Eric
-----Original Message----- From: Sandhya Sundaresan [mailto:[email protected]] Sent: Tuesday, August 4, 2015 1:19 PM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? RE: Where are the logs for the first set of regression tests? There are logs and diffs. I notice I added the instructions on the wiki only for the individual "pull request" automated jobs. I will add them for the official test results as well. Here they are : Click on the link in the daily build email : *https://jenkins02.trafodion.org/job/Check-Daily/48/* <https://jenkins02.trafodion.org/job/Check-Daily/48/> ==> Click on "Console Output" for any test suite you want to check ==> Around the 5th or 6th line you will see a link to detailed logs . Click that. 2015-08-04 08:30:13 Started by upstream project "Check-Daily" build number 48 2015-08-04 08:30:13 originally caused by: 2015-08-04 08:30:13 Started by timer 2015-08-04 08:30:13 [EnvInject] - Loading node environment variables. 2015-08-04 08:30:13 Building remotely on ahw22-172.16.0.75 (redhat_ahw2.2) in workspace /home/jenkins/workspace/core-regress-fullstack2-ahw2.2 2015-08-04 08:30:14 [core-regress-fullstack2-ahw2.2] $ /bin/sh /tmp/hudson2056470402644332566.sh 2015-08-04 08:30:14 Detailed logs: *http://logs.trafodion.org/Daily/2015-08-04_08-30-00/regress-fullstack2-ahw2.2/* <http://logs.trafodion.org/Daily/2015-08-04_08-30-00/regress-fullstack2-ahw2.2/> == > Navigate to sql-regress-log You will see a directory for the component test suite – eg “executor” and a full log – “Regress.log”. If you navigate to the component directory eg executor, you will see all the DIFFs and LOG files . Sandhya -----Original Message----- From: Eric Owhadi [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 11:06 AM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? Hi Sandhya, Thanks for the offer, I'll first do a re-run with the feature off and get familiar with the official results, that'll give me some learning opportunity :-). There is no "diff" file on the official results automatic daily run? So that it is easy to compare with the failure one get running the test manually? Eric -----Original Message----- From: Sandhya Sundaresan [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 12:50 PM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? Hi Eric, They are run on a nightly basis. Attached are the results form last night and previous night. As you can see 2 failures last night and both are being investigated. Otherwise we have been having good clean runs. So any other failures you see need to be investigated. If you can give me VNC access to your machine, I can take a quick look. I have documented how to read the official results on the wiki : https://wiki.trafodion.org/wiki/index.php/Contributor_Workflow#Checking_Test_Results Sandhya -----Original Message----- From: Eric Owhadi [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 10:40 AM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? Out of curiosity, anyone have recently run the regression test all on master branch? I am getting some failure, and I wonder if they are linked with my changes or if the current regression suit is having known temporary issues? Eric -----Original Message----- From: Dave Birdsall [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 11:26 AM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? Right. Start with the simple stuff I think. Eric's e-mail trail is valuable as it is fresh newbie experience; if we could answer Eric's questions on the wiki I think that would be a great start. -----Original Message----- From: Sandhya Sundaresan [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 9:24 AM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? Hi Eric, Dave, I can add a page to the wiki for this. We did have the steps documented in our old internal wiki for those starting on Trafodion for the first time. Also Roberta had a comprehensive document describing the whole regress environment back in MX days , I think. Not sure if anyone has that saved. Either way, for the wiki, perhaps a smaller page would be sufficient. Thanks Sandhya -----Original Message----- From: Eric Owhadi [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 9:03 AM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? I didn't want to ask, but indeed that would be fantastic and time saving. Also how to write and/or modify when tests are complaining, but we know that this is just because regression is based on delta, and cannot recognize legitimate expected delta (due to know feature we just implemented), and real regression. Eric -----Original Message----- From: Dave Birdsall [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 10:21 AM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? Do we have a page on our Apache Trafodion wiki that describes how to run the regression tests and how to interpret and deal with the results? If someone would like to volunteer to write such a page, it would be very useful to newbies, I think. -----Original Message----- From: Eric Owhadi [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 7:45 AM To: [email protected] Subject: RE: Where are the logs for the first set of regression tests? Thanks Suresh, that's it indeed :-). Was driving me crazy. Eric -----Original Message----- From: Suresh Subbiah [mailto:[email protected] <[email protected]>] Sent: Tuesday, August 4, 2015 9:41 AM To: [email protected] Subject: Re: Where are the logs for the first set of regression tests? Hi Eric, I think the "first" set of tests is the same as the last set. This is an artifact of how runallsb prints out results. At the end of every suite a summary of that suite is printed out (with no header what suite it refers to since we have been running that suite for the last X minutes). Then at the end of a run it concatenates the summary of every suite, this time with a header for each suite. So if you kick off regressions now and come back several hours later there will be a screen with a summary for all suites and then before the summary for core there will be list of results with no suite name. This is the same as the last suite in the summary (the hive test suite). Thanks Suresh On Tue, Aug 4, 2015 at 9:20 AM, Qifan Chen <[email protected]> wrote: > Hi Eric, > > The results should be in one of the directories under $rundir. For > example, under $rundir/core, the result for TEST001 (under > $scriptsdir/core) is LOG001. > > In my ~/.bashrc file, I have these two variables defined. > > export rundir=/mnt/qfc/rundir > > export scriptsdir=$trafodionRoot/sql/regress > > > After mkdir $rundir, make sure $rundir contains directories, one for > each category in $scriptsdir. Once it is setup, you can run > regressions as many times as you want :-) > > On Tue, Aug 4, 2015 at 9:01 AM, Eric Owhadi <[email protected]> wrote: > > > That may be an obvious question, but I can’t find the logs and diffs > > for the first set of test showing up in the result of the runallsb: > > > > I can see all the logs and diffs for categories like core, > > compGeneral > etc, > > but the first set of test labelled: > > > > TEST001 > > > > TEST002 > > > > TEST003 > > > > TEST004 > > > > TEST005 > > > > TEST006 > > > > TEST015 > > > > TEST017 > > > > TEST018 > > > > TEST020 > > > > > > > > are not associated with a category, and they don’t have a > > subdirectory > with > > categoryname where I can find the other test results. > > > > Thanks in advance for the help. > > > > > > > > As a side note, if a test fails because of the compare, but the > > delta is legit, what should one do? Update known diffs? Or set a new > > compare file? > > > > > > > > Eric > > > > > > -- > Regards, --Qifan >
