-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 4/29/11 10:47 AM, Patrick Crews wrote: > I'll have to dig into this, but something appears to be terribly borked > with tests related to the transaction log / innodb transaction log. > The tests are running, but we aren't seeing expected output : ( > You can always run this from the drizzle/tests dir: > ./dbqp --suite=transaction_log --debug --verbose > It is *quite* chatty, but it might give some insight. > > As for recreating the issues we saw on our FreeBSD machine, you can try > to create a drizzle branch somewhere with a ridiculously long path name > (something like 100 chars long) - that was the only case where I saw > things go awry (ie it was smooth as butter in my local repo) ; ) > > I'm stuck a conference now, so my bandwidth is limited...will see what I > can find out in a couple of days. > > Many, many thanks for helping out with this! > Patrick
Hi all, I have some new FreeBSD test run output for you here: http://paste.drizzle.org/show/494/ (dbqp transaction_log suite) http://paste.drizzle.org/show/495/ (test-all Makefile target) http://paste.drizzle.org/show/496/ (dbqp all suites) All of the tests passed, but perhaps someone with a better understanding of what's going on in the test process can tell if anything is amiss. However, there was a consistent problem with the "gmake test-all" run that I found a workaround for. When I first ran that target, I saw messages like this one every 4-5 tests: drizzle-test-run: WARNING: Forcing kill of process 92430 Sometimes the tests would continue (after killing said process, I assume), and other times the test execution would hang. Checking the process list, I always saw the drizzled process in the "STOP" state: # ps uaxww | grep drizzled | grep -v grep root 92430 0.0 3.0 289904 62464 1 T+ 12:38PM 0:00.51 /u1/usr/home/glarkin/FreeBSD/ports/databases/drizzle/work/drizzle7-2011.03.13/drizzled/.libs/drizzled - --no-defaults --default-storage-engine=innodb - --secure-file-priv=/u1/usr/home/glarkin/FreeBSD/ports/databases/drizzle/work/drizzle7-2011.03.13/tests/var - --tmpdir=/u1/usr/home/glarkin/FreeBSD/ports/databases/drizzle/work/drizzle7-2011.03.13/tests/var/tmp - --mysql-protocol.connect-timeout=60 --user=root - --pid-file=/u1/usr/home/glarkin/FreeBSD/ports/databases/drizzle/work/drizzle7-2011.03.13/tests/var/run/master.pid - --mysql-protocol.port=9110 --drizzle-protocol.port=9160 - --datadir=/u1/usr/home/glarkin/FreeBSD/ports/databases/drizzle/work/drizzle7-2011.03.13/tests/var/master-data - --mysql-unix-socket-protocol.path=/u1/usr/home/glarkin/FreeBSD/ports/databases/drizzle/work/drizzle7-2011.03.13/tests/var/master.sock - --server-id=1 --innodb.data-file-path=ibdata1:20M:autoextend - --sort-buffer-size=256K --max-heap-table-size=1M #--plugin-add=catalog - --core-file I tried sending SIGCONT to the process, and that allowed the process to exit and test execution to proceed: ... drizzle-test-run: WARNING: Forcing kill of process 92430 transaction_log.alter [ pass ] 838 transaction_log.auto_commit [ pass ] 183 transaction_log.blob [ pass ] 188 transaction_log.create_select [ pass ] 190 transaction_log.create_table [ pass ] 1077 ... ... I eventually cobbled together a while loop in another window to check for stopped drizzled processes every 5 seconds and send SIGCONT to them. After I did that, the whole test suite completed successfully without any other interaction. Does anyone know why the processes stop, and is this related to the other hangs that were originally reported a couple of weeks ago? Thank you, Greg -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk3AMlgACgkQ0sRouByUApAXSgCgtgLi3nVbu02Qv213dK8pT0CZ zMYAn1k6K6W5Y6aUjtMrCFDp2HJ0ENBw =QDk6 -----END PGP SIGNATURE----- _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
