Re: fairywren timeout failures on the pg_amcheck/005_opclass_damage test

On 2024-07-25 Th 8:00 AM, Alexander Lakhin wrote:
> Hello hackers,
>
> Please take a look at today's failure [1], that raises several questions
> at once:
> 117/244 postgresql:pg_upgrade / pg_upgrade/002_pg_upgrade 
> TIMEOUT        3001.48s   exit status 1
> 180/244 postgresql:pg_amcheck / pg_amcheck/005_opclass_damage 
> TIMEOUT        3001.43s   exit status 1
>
> Ok:                 227
> Expected Fail:      0
> Fail:               0
> Unexpected Pass:    0
> Skipped:            15
> Timeout:            2
>
> But looking at the previous test run [2], marked 'OK', I can see almost
> the same:
> 115/244 postgresql:pg_upgrade / pg_upgrade/002_pg_upgrade 
> TIMEOUT        3001.54s   exit status 1
> 176/244 postgresql:pg_amcheck / pg_amcheck/005_opclass_damage 
> TIMEOUT        3001.26s   exit status 1
>
> Ok:                 227
> Expected Fail:      0
> Fail:               0
> Unexpected Pass:    0
> Skipped:            15
> Timeout:            2
>
> So it's not clear to me, why is the latter test run considered failed
> unlike the former?


Yesterday I changed the way we detect failure in the buildfarm client, 
and pushed that to fairywren (and a few more of my animals). Previously 
it did not consider a timeout to be a failure, but now it does. I'm 
going to release that soon.


cheers


andrew


--
Andrew Dunstan
EDB: https://www.enterprisedb.com