Thread: Can we include capturing logs of pgdata/pg_upgrade_output.d/*/log in buildfarm
Hi, Recently there have been few upgrade tap test failures in buildfarm like in [1] & [2]. Analysing these failures requires the log files that are getting generated from src/bin/pg_upgrade at the following locations: tmp_check/*/pgdata/pg_upgrade_output.d/*/*.txt - e.g. tmp_check/t_004_subscription_new_sub1_data/pgdata/pg_upgrade_output.d/20240214T052229.045/subs_invalid.txt tmp_check/*/pgdata/pg_upgrade_output.d/*/*/*.log - e.g. tmp_check/t_004_subscription_new_sub1_data/pgdata/pg_upgrade_output.d/20240214T052229.045/log/pg_upgrade_server.log First regex is the testname_clusterinstance_data, second regex is the timestamp used for pg_upgrade, third regex is for the text files generated by pg_upgrade and fourth regex is for the log files generated by pg_upgrade. Can we include these log files also in the buildfarm? [1] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2024-02-10%2007%3A03%3A10 [2] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2023-12-07%2003%3A56%3A20 Regards, Vignesh
Re: Can we include capturing logs of pgdata/pg_upgrade_output.d/*/log in buildfarm
From
Michael Paquier
Date:
On Wed, Feb 14, 2024 at 03:51:08PM +0530, vignesh C wrote: > First regex is the testname_clusterinstance_data, second regex is the > timestamp used for pg_upgrade, third regex is for the text files > generated by pg_upgrade and fourth regex is for the log files > generated by pg_upgrade. > > Can we include these log files also in the buildfarm? > > [1] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2024-02-10%2007%3A03%3A10 > [2] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2023-12-07%2003%3A56%3A20 Indeed, these lack some patterns. Why not sending a pull request around [1] to get more patterns covered? [1]: https://github.com/PGBuildFarm/client-code/blob/main/PGBuild/Modules/TestUpgrade.pm -- Michael
Attachment
Re: Can we include capturing logs of pgdata/pg_upgrade_output.d/*/log in buildfarm
From
vignesh C
Date:
On Thu, 15 Feb 2024 at 07:24, Michael Paquier <michael@paquier.xyz> wrote: > > On Wed, Feb 14, 2024 at 03:51:08PM +0530, vignesh C wrote: > > First regex is the testname_clusterinstance_data, second regex is the > > timestamp used for pg_upgrade, third regex is for the text files > > generated by pg_upgrade and fourth regex is for the log files > > generated by pg_upgrade. > > > > Can we include these log files also in the buildfarm? > > > > [1] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2024-02-10%2007%3A03%3A10 > > [2] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2023-12-07%2003%3A56%3A20 > > Indeed, these lack some patterns. Why not sending a pull request > around [1] to get more patterns covered? I have added a few more patterns to include the pg_upgrade generated files. The attached patch has the changes for the same. Adding Andrew also to get his thoughts on this. Regards, Vignesh
Attachment
Re: Can we include capturing logs of pgdata/pg_upgrade_output.d/*/log in buildfarm
From
vignesh C
Date:
On Thu, 15 Feb 2024 at 08:36, vignesh C <vignesh21@gmail.com> wrote: > > On Thu, 15 Feb 2024 at 07:24, Michael Paquier <michael@paquier.xyz> wrote: > > > > On Wed, Feb 14, 2024 at 03:51:08PM +0530, vignesh C wrote: > > > First regex is the testname_clusterinstance_data, second regex is the > > > timestamp used for pg_upgrade, third regex is for the text files > > > generated by pg_upgrade and fourth regex is for the log files > > > generated by pg_upgrade. > > > > > > Can we include these log files also in the buildfarm? > > > > > > [1] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2024-02-10%2007%3A03%3A10 > > > [2] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2023-12-07%2003%3A56%3A20 > > > > Indeed, these lack some patterns. Why not sending a pull request > > around [1] to get more patterns covered? > > I have added a few more patterns to include the pg_upgrade generated > files. The attached patch has the changes for the same. > Adding Andrew also to get his thoughts on this. I have added the following commitfest entry for this: https://commitfest.postgresql.org/47/4850/ Regards, Vignesh
Re: Can we include capturing logs of pgdata/pg_upgrade_output.d/*/log in buildfarm
From
Andrew Dunstan
Date:
On 2024-02-25 Su 11:18, vignesh C wrote: > On Thu, 15 Feb 2024 at 08:36, vignesh C <vignesh21@gmail.com> wrote: >> On Thu, 15 Feb 2024 at 07:24, Michael Paquier <michael@paquier.xyz> wrote: >>> On Wed, Feb 14, 2024 at 03:51:08PM +0530, vignesh C wrote: >>>> First regex is the testname_clusterinstance_data, second regex is the >>>> timestamp used for pg_upgrade, third regex is for the text files >>>> generated by pg_upgrade and fourth regex is for the log files >>>> generated by pg_upgrade. >>>> >>>> Can we include these log files also in the buildfarm? >>>> >>>> [1] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2024-02-10%2007%3A03%3A10 >>>> [2] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2023-12-07%2003%3A56%3A20 >>> Indeed, these lack some patterns. Why not sending a pull request >>> around [1] to get more patterns covered? >> I have added a few more patterns to include the pg_upgrade generated >> files. The attached patch has the changes for the same. >> Adding Andrew also to get his thoughts on this. > I have added the following commitfest entry for this: > https://commitfest.postgresql.org/47/4850/ > Buildfarm code patches do not belong in the Commitfest, I have marked the item as rejected. You can send me patches directly or add a PR to the buildfarm's github repo. In this case the issue on drongo was a typo, the fix for which I had forgotten to propagate back in December. Note that the buildfarm's TestUpgrade.pm module is only used for branches < 15. For branches >= 15 we run the standard TAP test and this module does nothing. More generally, the collection of logs etc. for pg_upgrade will improve with the next release, which will be soon after I return from a vacation in about 2 weeks - experience shows that making releases just before a vacation is not a good idea :-) cheers andrew -- Andrew Dunstan EDB: https://www.enterprisedb.com
Re: Can we include capturing logs of pgdata/pg_upgrade_output.d/*/log in buildfarm
From
vignesh C
Date:
On Mon, 26 Feb 2024 at 10:57, Andrew Dunstan <andrew@dunslane.net> wrote: > > > On 2024-02-25 Su 11:18, vignesh C wrote: > > On Thu, 15 Feb 2024 at 08:36, vignesh C <vignesh21@gmail.com> wrote: > >> On Thu, 15 Feb 2024 at 07:24, Michael Paquier <michael@paquier.xyz> wrote: > >>> On Wed, Feb 14, 2024 at 03:51:08PM +0530, vignesh C wrote: > >>>> First regex is the testname_clusterinstance_data, second regex is the > >>>> timestamp used for pg_upgrade, third regex is for the text files > >>>> generated by pg_upgrade and fourth regex is for the log files > >>>> generated by pg_upgrade. > >>>> > >>>> Can we include these log files also in the buildfarm? > >>>> > >>>> [1] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2024-02-10%2007%3A03%3A10 > >>>> [2] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=drongo&dt=2023-12-07%2003%3A56%3A20 > >>> Indeed, these lack some patterns. Why not sending a pull request > >>> around [1] to get more patterns covered? > >> I have added a few more patterns to include the pg_upgrade generated > >> files. The attached patch has the changes for the same. > >> Adding Andrew also to get his thoughts on this. > > I have added the following commitfest entry for this: > > https://commitfest.postgresql.org/47/4850/ > > > > Buildfarm code patches do not belong in the Commitfest, I have marked > the item as rejected. You can send me patches directly or add a PR to > the buildfarm's github repo. Ok, I will send over the patch directly for the required things. > > In this case the issue on drongo was a typo, the fix for which I had > forgotten to propagate back in December. Note that the buildfarm's > TestUpgrade.pm module is only used for branches < 15. For branches >= 15 > we run the standard TAP test and this module does nothing. > > More generally, the collection of logs etc. for pg_upgrade will improve > with the next release, which will be soon after I return from a vacation > in about 2 weeks - experience shows that making releases just before a > vacation is not a good idea :-) Thanks, that will be helpful. Regards, Vignesh