Re: Failed to make v10 branch: gitmaster push hook needs adjustment

Andrew, Tom, Andrew,

* Andres Freund (andres(at)anarazel(dot)de) wrote:
> On 2017-08-14 23:28:21 +0000, Stephen Frost wrote:
> > On Mon, Aug 14, 2017 at 19:21 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > > Andres Freund <andres(at)anarazel(dot)de> writes:
> > > > On 2017-08-14 19:09:23 -0400, Andrew Dunstan wrote:
> > > >> I am unfortunately away for about a week. Perhaps Stephen Frost can
> > > help with any emergency patches to the server
> > >
> > > > Stephen? Any chance to relax the restriction here? It'd be somewhat bad
> > > > if we didn't have pg 10 buildfarm coverage over the next few days...
> > >
> > > In addition to just allowing submissions for v10, it'd be great if
> > > we could get the list of active branches for run_branches.pl updated,
> > > so that the critters with non-manually-maintained branch lists start
> > > doing the right thing.
> > >
> > > (Maybe that's the exact same thing, in which case nevermind.)
> >
> >
> > I can certainly patch servers if needed. I'll be back to a keyboard in
> > about 10m, tho I am at a loss as to what exactly needs doing here..?
> >
> > I'll try to find the rest of the thread when I reach a real computer.
>
> Based on a quick look it seems the error is triggered by
> https://github.com/PGBuildFarm/server-code/blob/master/cgi-bin/pgstatus.pl
>
> There appear to be two levels of checks here. First:
>
> if (open($brhandle,"../htdocs/branches_of_interest.txt"))
> {
> my @branches_of_interest = <$brhandle>;
> close($brhandle);
> chomp(@branches_of_interest);
> unless (grep {$_ eq $branch} @branches_of_interest)
> {
> print
> "Status: 492 bad branch parameter $branch\n",
> "Content-Type: text/plain\n\n",
> "bad branch parameter $branch\n";
> exit;
> }
> }
>
> so that file needs to be updated.

Done.

> Secondly there's:
> unless ($branch =~ /^(HEAD|REL\d+_\d+_STABLE)$/)
> {
> print
> "Status: 492 bad branch parameter $branch\nContent-Type: text/plain\n\n",
> "bad branch parameter $branch\n";
> exit;
> }
>
> which needs its regex adjusted... Based on a quick look preliminarily it
> seems to be ok to just make the first \d+ optional (i.e. (\d+)?).

Done.

Let's see what happens... I'm around for the next few hours if things
go south.

Thanks!

Stephen