On 06/24/2011 08:09 AM, Olaf van der Spek wrote:
On Fri, Jun 24, 2011 at 2:06 PM, Andrew Hutchings
<[email protected]> wrote:
True, but testing is important.
I know, that's why I'm looking for ways to speed it up.
1) It's a lot of tests - it is a complicated system.
2) It's better than it has been - used to be 30+ minutes
3) parallel execution is coming to dbqp eventually
4) If waiting 10 minutes is absolutely killing you, I would recommend
trying to patch dbqp as test-run.pl will be going away before long.
You can code another branch whilst it is running.
That's not ideal.
It is what the rest of us do.
What speedup can be expected from libeatmydata?
Can't we just pass a switch to Drizzle to disable syncs (for testing)
or would that be more complicated?
I have a patch to this for dbqp that I'll be proposing to merge today.
If you use ./dbqp --libeatmydata [--libeatmydata-path-...], it will just
use libeatmydata (if installed / detected) for spinning up servers
Additionally, make test-dbqp uses shared mem by default, so make test =
20110623-135133 INFO Test execution complete in 490 seconds
20110623-135133 INFO Summary report:
20110623-135133 INFO Executed 566/566 test cases, 100.00 percent
20110623-135133 INFO STATUS: PASS, 566/566 test cases, 100.00 percent
executed
For me it is around a 10x speedup. If you are using a BBU RAID array
then there probably won't be much speedup.
What time does make test take for you?
Olaf
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
