[Gluster-devel] geo-rep regression tests take *ages*?

[Aravinda]

I sent patch with some improvements, but it can still be improved.

http://review.gluster.org/#/c/10340/

Changes done today are,
1. Reduced sleep intervals
2. Reduced Timeouts
3. Reduced sample data size for testing

/tests/geo-rep/georep-rsync-changelog.t completed in 7mins in my machine.

I think all the hybrid test cases(georep-rsync-hybrid.t and 
georep-tarssh-hybrid.t) we can disable, since these are very old test 
cases it assumes on every geo-rep restart it uses Hybrid crawl. But it 
is not true, now geo-rep uses History Changelogs when geo-rep restarted.

--
regards
Aravinda

On 04/22/2015 08:17 PM, Jeff Darcy wrote:
>> Just noticed something a bit weird on the regression tests
>> for CentOS 6.x:
>>
>>    [13:28:44] ./tests/features/weighted-rebalance.t
>>    ....................................................... ok       23 s
>>    [13:46:50] ./tests/geo-rep/georep-rsync-changelog.t
>>    .................................................... ok     1086 s
>>    [14:06:53] ./tests/geo-rep/georep-rsync-hybrid.t
>>    ....................................................... ok     1203 s
>>    [14:08:36] ./tests/geo-rep/georep-setup.t
>>    .............................................................. ok      103
>>    s
>>    [14:26:35] ./tests/geo-rep/georep-tarssh-changelog.t
>>    ................................................... ok     1079 s
>>
>> That's on:
>>
>>    http://build.gluster.org/job/rackspace-regression-2GB-triggered/7285/console
>>
>> Those 3x 1000+ second regression tests are adding 56+minutes to the total
>> regression test time.
>>
>> That's not how it should be is it?
> I reported this to the geo-rep team last week.  Apparently it's a low
> priority for them, and the tests are so convoluted that I haven't figured
> out how to fix it myself.
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-devel