big test is failing with dbqp

[Lee Bieber]

"big" test started failing. it passes using the old test runner, but started failing today with the new test runner. Do we perhaps have to run this test with --no-shm? But if so, what changed to cause that as it was passing in the past.

http://jenkins.drizzle.org/view/Drizzle-build/job/drizzle-build-big-test/100/console

10 Feb 2011 20:36:43 INFO: STATUS: 1
10 Feb 2011 20:38:49 : ================================================================================
10 Feb 2011 20:38:49 : TEST NAME [ RESULT ] TIME (ms)
10 Feb 2011 20:38:49 : ================================================================================
10 Feb 2011 20:38:49 : big.archive-big [ fail ] 125506
10 Feb 2011 20:38:49 : drizzletest: At line 22: query 'INSERT INTO t1 SELECT * FROM t1' failed: 1030: Got error -1 from storage engine
10 Feb 2011 20:38:49 :
10 Feb 2011 20:38:49 : The result from queries just before the failure was:
10 Feb 2011 20:38:49 : CREATE TABLE t1(a BLOB) ENGINE=ARCHIVE;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 : INSERT INTO t1 SELECT * FROM t1;
10 Feb 2011 20:38:49 :
10 Feb 2011 20:38:49 : More results from queries before failure can be found in /home/hudson/hudson/workspace/drizzle-build-big-test/tests/dbqp_work/testbot0/server0/var/log/archive-big.log
10 Feb 2011 20:38:49 :
10 Feb 2011 20:38:49 ERROR: Failed test. Use --force to execute beyond the first test failure
10 Feb 2011 20:38:52 : ================================================================================
10 Feb 2011 20:38:52 INFO: Test execution complete in 128 seconds
10 Feb 2011 20:38:52 INFO: Summary report:
10 Feb 2011 20:38:52 INFO: Executed 1/4 test cases, 25.00 percent
10 Feb 2011 20:38:52 INFO: STATUS: FAIL, 1/1 test cases, 100.00 percent executed
10 Feb 2011 20:38:52 INFO: FAIL tests: big.archive-big
10 Feb 2011 20:38:52 INFO: Spent 125 / 128 seconds on: TEST(s)
10 Feb 2011 20:38:52 INFO: Test execution complete
10 Feb 2011 20:38:52 INFO: Stopping all running servers...

[Stewart Smith (stewart)]

On Fri, 11 Feb 2011 20:35:26 -0000, Lee Bieber <email address hidden> wrote:
> "big" test started failing. it passes using the old test runner, but
> started failing today with the new test runner. Do we perhaps have to
> run this test with --no-shm? But if so, what changed to cause that as it
> was passing in the past.

I'm pretty sure it's because of running on 8GB or less machines with
/dev/shm for datadir. I had to remove 3.9GB of stuff from /dev/shm when
it failed on my machine.... and /dev/shm is limited to 1/2 of physical
memory by default.

We should just use libeatmydata instead.

--
Stewart Smith

[Patrick Crews (patrick-crews)]

As we have removed the archive engine, this is a moot point. Will investigate switching over to use of libeatmydata for server speed-ups vs. tmpfs in the future. Closing the bug as won't fix