[Gluster-devel] Auto-retry failed tests

Justin Clift justin at gluster.org
Thu Apr 16 15:08:10 UTC 2015


On 16 Apr 2015, at 05:28, Emmanuel Dreyfus <manu at netbsd.org> wrote:
> Hi
> 
> We all know regression spurious failures are a problem. In order to
> minimize their impact, NetBSD regression restart the whole test suite in
> case of error so that spurious failures do not cause an undeserved
> verified=-1 vote to be cast.
> 
> This takes time, and as a consequence the netbsd7_regression backlog
> gets huge in the afternoon.
> 
> I proposed this change to improve the situation: modify run-tests.sh to
> retry only the failed tests. That behavior is off by default and can be
> enabled using run-tests.sh -r 
> http://review.gluster.org/10128/
> 
> Having that one merged would not change anything to the way Linux
> regression is run right now, and it would let me make the NetBSD
> regression much faster (until the day spurious regressions are fixed).

The concept sounds really useful. :)

+ Justin

--
GlusterFS - http://www.gluster.org

An open source, distributed file system scaling to several
petabytes, and handling thousands of clients.

My personal twitter: twitter.com/realjustinclift



More information about the Gluster-devel mailing list