Обсуждение: Initdb-cs_CZ.WIN-1250 buildfarm failures

Buildfarm members magpie, treepie and fulmar went absent on 2014-10-29.  Since
returning on 2014-11-16, they have consistently failed with 'initdb: invalid
locale name "cs_CZ.WIN-1250"'.  No commits in that period readily explain a
regression in this area.  Did the underlying system configurations change?

Thanks,
nm

On Sat, Dec 20, 2014 at 05:14:03PM +0100, CSPUG wrote:
> On 20.12.2014 07:39, Noah Misch wrote:
> > Buildfarm members magpie, treepie and fulmar went absent on
> > 2014-10-29. Since returning on 2014-11-16, they have consistently
> > failed with 'initdb: invalid locale name "cs_CZ.WIN-1250"'. No
> > commits in that period readily explain a regression in this area. Did
> > the underlying system configurations change?
> 
> I'm pretty sure it's because of broken locales at the system level. It was
> working fine, and I haven't done any substantial changes to the system
> except for occasional "yum update", so I'm unaware of what went wong :(
> 
> The issue looks like this:
> 
> # locale -a | grep en_US
> en_US
> en_US.iso88591
> en_US.iso885915
> en_US.utf8
> 
> # locale en_US
> locale: unknown name "en_US"

The NAME argument to the "locale" tool is something like LC_PAPER, not a
locale name.  Use "LANG=en_US locale LC_NUMERIC" to test locale loading.

> The only reasons I can think of is that some of the updates required
> a reboot, and I haven't done that because that would kill all the VMs
> running on that HW, including the one with CLOBBER_CACHE_RECURSIVE
> tests. And that'd throw away tests running for ~3 months.
> 
> I've disabled the three animals (magpie, fulmar, treepie) for now, because
> there's no point in running the tests until the locale issues are fixed. If
> anyone has an idea of what might be wrong, let me know.

Those animals have been successfully completing initdb for several locales,
including en_US, before failing at cs_CZ.WIN-1250.  You could disable just the
cs_CZ.WIN-1250 steps.  A CentOS 6.6 system here also lacks such a locale:

$ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
locale: Cannot set LC_CTYPE to default locale: No such file or directory
locale: Cannot set LC_MESSAGES to default locale: No such file or directory
locale: Cannot set LC_ALL to default locale: No such file or directory
.

-1
46
0
ANSI_X3.4-1968
$ locale -a|grep cs_
cs_CZ
cs_CZ.iso88592
cs_CZ.utf8

Perhaps an update made the system stricter about rejecting unknown locales.

Noah Misch <noah@leadboat.com> writes:
> On Sat, Dec 20, 2014 at 05:14:03PM +0100, CSPUG wrote:
>> On 20.12.2014 07:39, Noah Misch wrote:
>>> Buildfarm members magpie, treepie and fulmar went absent on
>>> 2014-10-29. Since returning on 2014-11-16, they have consistently
>>> failed with 'initdb: invalid locale name "cs_CZ.WIN-1250"'. No
>>> commits in that period readily explain a regression in this area. Did
>>> the underlying system configurations change?

That's what I'd been assuming, but I had not got round to inquiring.

>> I'm pretty sure it's because of broken locales at the system level. It was
>> working fine, and I haven't done any substantial changes to the system
>> except for occasional "yum update", so I'm unaware of what went wong :(
>> 
>> The only reasons I can think of is that some of the updates required
>> a reboot, and I haven't done that because that would kill all the VMs
>> running on that HW, including the one with CLOBBER_CACHE_RECURSIVE
>> tests. And that'd throw away tests running for ~3 months.
>> 
>> I've disabled the three animals (magpie, fulmar, treepie) for now, because
>> there's no point in running the tests until the locale issues are fixed. If
>> anyone has an idea of what might be wrong, let me know.

> Those animals have been successfully completing initdb for several locales,
> including en_US, before failing at cs_CZ.WIN-1250.  You could disable just the
> cs_CZ.WIN-1250 steps.  A CentOS 6.6 system here also lacks such a locale:

> $ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
> locale: Cannot set LC_CTYPE to default locale: No such file or directory
> locale: Cannot set LC_MESSAGES to default locale: No such file or directory
> locale: Cannot set LC_ALL to default locale: No such file or directory

FWIW, an actual RHEL 6.6 system doesn't like that either; at least not
with a fairly minimal set of locales installed.

> Perhaps an update made the system stricter about rejecting unknown locales.

Red Hat released RHEL 6.6 on Oct 14.  I am not sure how long it took the
CentOS crew to follow suit, but it's well within the realm of plausibility
that when these buildfarm critters went offline it was associated with a
system update from 6.5 to 6.6 ... which was a *major* update.  (On my
machine, something like 480 out of 1500 packages were updated.)  It's easy
to believe that either the glibc locale code is now stricter, or that Red
Hat reshuffled the packaging a bit and the missing locale is now in a
new package that didn't get installed.

Personally I'd not have tried to run such a system without rebooting ...
it's unlikely that not rebooting has anything directly to do with this
problem, but I would not be surprised by issues elsewhere.
        regards, tom lane

On 20.12.2014 18:13, Pavel Stehule wrote:
> 
> 2014-12-20 17:48 GMT+01:00 Noah Misch <noah@leadboat.com
>
>     $ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
> 
> 
> It is Microsoft encoding, - it is not available on Linux

Not true. It is available on Linux, and the regression tests were
running with it for a long time (essentially from the moment magpie was
added to the buildfarm).

Tomas

Tomas Vondra <tv@fuzzy.cz> writes:
> On 20.12.2014 18:13, Pavel Stehule wrote:
>> It is Microsoft encoding, - it is not available on Linux

> Not true. It is available on Linux, and the regression tests were
> running with it for a long time (essentially from the moment magpie was
> added to the buildfarm).

Well, you had it at one time, but it sure doesn't seem to be there now.

I poked around in the RHEL 6.6 release notes, and found this possibly
relevant tidbit:
 mingw component, BZ#1063396
 Following the deprecation of Matahari packages in Red Hat Enterprise Linux 6.3, at which time the mingw packages were
notedas deprecated, and the subsequent removal of Matahari packages from Red Hat Enterprise Linux 6.4, the mingw
packagesare now being removed from Red Hat Enterprise Linux 6.6.
 
 The mingw packages will no longer be shipped in future Red Hat Enterprise Linux 6 minor releases, nor will they
receivesecurity-related updates. Consequently, users are advised to uninstall any earlier releases of the mingw
packagesfrom their Red Hat Enterprise Linux 6 systems.
 

It seems plausible that a WIN-1250 locale would have been something
that would've been supplied by mingw rather than being part of either
glibc-common or any official locale package.  I'm not sure how that
translates to it actively disappearing from your machine --- as the note
says, you're "advised to uninstall" mingw, but I don't think the 6.6
update would have removed it automatically.  Still, the outlines of a
theory are starting to come into focus.

Immediate recommendation is to restart the buildfarm critters with the
missing locale removed from their configurations.  If you can find the
locale again somewhere, by all means re-enable it, but better less testing
than no testing in the meantime.
        regards, tom lane

On 12/20/2014 12:32 PM, Tom Lane wrote:

>
> Immediate recommendation is to restart the buildfarm critters with the
> missing locale removed from their configurations.  If you can find the
> locale again somewhere, by all means re-enable it, but better less testing
> than no testing in the meantime.
>
>             


Yeah, all they need to do is alter the locales setting in the config 
file. If they want to force new runs, they can use 
<https://wiki.postgresql.org/wiki/PostgreSQL_Buildfarm_Howto#Tips_and_Tricks>


cheers

andrew

On 20.12.2014 17:48, Noah Misch wrote:
> On Sat, Dec 20, 2014 at 05:14:03PM +0100, CSPUG wrote:
>> On 20.12.2014 07:39, Noah Misch wrote:
>>> Buildfarm members magpie, treepie and fulmar went absent on
>>> 2014-10-29. Since returning on 2014-11-16, they have consistently
>>> failed with 'initdb: invalid locale name "cs_CZ.WIN-1250"'. No
>>> commits in that period readily explain a regression in this area. Did
>>> the underlying system configurations change?
>>
>> I'm pretty sure it's because of broken locales at the system level. It was
>> working fine, and I haven't done any substantial changes to the system
>> except for occasional "yum update", so I'm unaware of what went wong :(
>>
>> The issue looks like this:
>>
>> # locale -a | grep en_US
>> en_US
>> en_US.iso88591
>> en_US.iso885915
>> en_US.utf8
>>
>> # locale en_US
>> locale: unknown name "en_US"
>
> The NAME argument to the "locale" tool is something like LC_PAPER, not a
> locale name.  Use "LANG=en_US locale LC_NUMERIC" to test locale loading.

D'oh! That kinda explains the strange `locale` failures, of course ...

But still, the three animals (magpie, fulmar, treepie) are all running
with these locales:

   C en_US
   cs_CZ.UTF-8 cs_CZ.ISO-8859-2 cs_CZ.WIN-1250
   sk_SK.UTF-8 sk_SK.ISO-8859-2 sk_SK.WIN-1250

and the only one missing (so the 'locale LC_NUMERIC' failed) was
sk_SK.WIN-1250. I've fixed that, and now all the locales work fine.

That however does not explain why the tests are failing with
cs_CZ.WIN-1250 because that locale was working for some time.

>
>> The only reasons I can think of is that some of the updates required
>> a reboot, and I haven't done that because that would kill all the VMs
>> running on that HW, including the one with CLOBBER_CACHE_RECURSIVE
>> tests. And that'd throw away tests running for ~3 months.
>>
>> I've disabled the three animals (magpie, fulmar, treepie) for now, because
>> there's no point in running the tests until the locale issues are fixed. If
>> anyone has an idea of what might be wrong, let me know.
>
> Those animals have been successfully completing initdb for several locales,
> including en_US, before failing at cs_CZ.WIN-1250.  You could disable just the
> cs_CZ.WIN-1250 steps.  A CentOS 6.6 system here also lacks such a locale:
>
> $ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
> locale: Cannot set LC_CTYPE to default locale: No such file or directory
> locale: Cannot set LC_MESSAGES to default locale: No such file or directory
> locale: Cannot set LC_ALL to default locale: No such file or directory
> .
>
> -1
> 46
> 0
> ANSI_X3.4-1968
> $ locale -a|grep cs_
> cs_CZ
> cs_CZ.iso88592
> cs_CZ.utf8
>
> Perhaps an update made the system stricter about rejecting unknown locales.

I believe the locale system (at the OS level) works just like before. I
remember I had to manually create the locales while initially setting up
the animals. Then, ~2 months ago something happened (I asssume a yum
update) and some of the locales disappeared. But I have recreated them,
except for sk_SK.WIN-1250. But the tests fail because of cs_CZ.WIN-1250
which does exist.

Attached is a log of

for l in cs_CZ.UTF-8 sk_SK.UTF-8 cs_CZ.ISO-8859-2 sk_SK.ISO-8859-2
cs_CZ.WIN-1250 sk_SK.WIN-1250; do
  echo $l; LANG=$l locale LC_NUMERIC;
done;

showing that all the locales exist. However when I tried to initialize a
cluster with cs_CZ.WIN-1250, I got an error like this:

[pgbuild@regular-builds ~]$ locale
LANG=cs_CZ.WIN-1250
LC_CTYPE="cs_CZ.WIN-1250"
LC_NUMERIC="cs_CZ.WIN-1250"
LC_TIME="cs_CZ.WIN-1250"
LC_COLLATE="cs_CZ.WIN-1250"
LC_MONETARY="cs_CZ.WIN-1250"
LC_MESSAGES="cs_CZ.WIN-1250"
LC_PAPER="cs_CZ.WIN-1250"
LC_NAME="cs_CZ.WIN-1250"
LC_ADDRESS="cs_CZ.WIN-1250"
LC_TELEPHONE="cs_CZ.WIN-1250"
LC_MEASUREMENT="cs_CZ.WIN-1250"
LC_IDENTIFICATION="cs_CZ.WIN-1250"
LC_ALL=
[pgbuild@regular-builds ~]$ pg_ctl -D tmp-data init
The files belonging to this database system will be owned by user "pgbuild".
This user must also own the server process.

The database cluster will be initialized with locale "cs_CZ.WIN-1250".
could not determine encoding for locale "cs_CZ.WIN-1250": codeset is
"ANSI_X3.4-1968"
initdb: could not find suitable encoding for locale "cs_CZ.WIN-1250"
Rerun initdb with the -E option.
Try "initdb --help" for more information.
pg_ctl: database system initialization failed

So apparently the locale does exist, but we're unable to cope with it
for some reason ...

Tomas

On 20.12.2014 18:32, Tom Lane wrote:
> Tomas Vondra <tv@fuzzy.cz> writes:
>> On 20.12.2014 18:13, Pavel Stehule wrote:
>>> It is Microsoft encoding, - it is not available on Linux
> 
>> Not true. It is available on Linux, and the regression tests were 
>> running with it for a long time (essentially from the moment magpie
>> was added to the buildfarm).
> 
> Well, you had it at one time, but it sure doesn't seem to be there now.
>
> I poked around in the RHEL 6.6 release notes, and found this possibly
> relevant tidbit:

Ah, apparently I upgraded this VM to 6.6 - that wasn't exactly planned.

> 
>      mingw component, BZ#1063396
> 
>      Following the deprecation of Matahari packages in Red Hat
>      Enterprise Linux 6.3, at which time the mingw packages were noted
>      as deprecated, and the subsequent removal of Matahari packages
>      from Red Hat Enterprise Linux 6.4, the mingw packages are now
>      being removed from Red Hat Enterprise Linux 6.6.
> 
>      The mingw packages will no longer be shipped in future Red Hat
>      Enterprise Linux 6 minor releases, nor will they receive
>      security-related updates. Consequently, users are advised to
>      uninstall any earlier releases of the mingw packages from their
>      Red Hat Enterprise Linux 6 systems.
> 
> It seems plausible that a WIN-1250 locale would have been something 
> that would've been supplied by mingw rather than being part of
> either glibc-common or any official locale package. I'm not sure how
> that translates to it actively disappearing from your machine --- as
> the note says, you're "advised to uninstall" mingw, but I don't think
> the 6.6 update would have removed it automatically. Still, the
> outlines of a theory are starting to come into focus.

I don't think so. As I mentioned in the previous post, I still can
create the locale, but initdb still fails with an error about
ANSI_X3.4-1968 encoding.

But clearly, something changed between RH 6.5 and 6.6, because on 6.5 I
get this:

$ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
,
�
3;3
44
160
CP1250

while on 6.6 I get this:

$ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
,

3;3
44
160
ANSI_X3.4-1968

So yeah, the locale definition did change a bit - the  encoding is
different, and we can't cope with it.

> Immediate recommendation is to restart the buildfarm critters with
> the missing locale removed from their configurations. If you can find
> the locale again somewhere, by all means re-enable it, but better
> less testing than no testing in the meantime.

I've removed cs_CZ.WIN-1250 and sk_SK.WIN-1250 from the config. Let's
see how that works.


Tomas

Tomas Vondra <tv@fuzzy.cz> writes:
> I believe the locale system (at the OS level) works just like before. I
> remember I had to manually create the locales while initially setting up
> the animals. Then, ~2 months ago something happened (I asssume a yum
> update) and some of the locales disappeared. But I have recreated them,
> except for sk_SK.WIN-1250. But the tests fail because of cs_CZ.WIN-1250
> which does exist.

I am betting that you recreated them differently from before.

> However when I tried to initialize a
> cluster with cs_CZ.WIN-1250, I got an error like this:

> [pgbuild@regular-builds ~]$ pg_ctl -D tmp-data init
> The files belonging to this database system will be owned by user "pgbuild".
> This user must also own the server process.

> The database cluster will be initialized with locale "cs_CZ.WIN-1250".
> could not determine encoding for locale "cs_CZ.WIN-1250": codeset is
> "ANSI_X3.4-1968"
> initdb: could not find suitable encoding for locale "cs_CZ.WIN-1250"

Locale cs_CZ.WIN-1250 is evidently marked with a codeset property of
"ANSI_X3.4-1968" (which means old-school US-ASCII).  That's certainly
wrong.  I believe the correct thing would be "CP1250".
        regards, tom lane

Tomas Vondra <tv@fuzzy.cz> writes:
> But clearly, something changed between RH 6.5 and 6.6, because on 6.5 I
> get this:

> $ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
> ,
> �
> 3;3
> 44
> 160
> CP1250

> while on 6.6 I get this:

> $ LANG=cs_CZ.WIN-1250 locale LC_NUMERIC
> ,

> 3;3
> 44
> 160
> ANSI_X3.4-1968

That's certainly broken.  The entire point of having a cs_CZ.WIN-1250
locale (as opposed to cs_CZ.something-else) would be to specify a
codeset corresponding to WIN-1250.  Our code recognizes the spelling
"CP1250" for that.  It's possible there are other spellings we should
recognize, but "ANSI_X3.4-1968" is certainly not one.  As I said, that
just means ASCII, so it's completely useless for determining which
ASCII-superset encoding is wanted.
        regards, tom lane

On 20.12.2014 19:05, Tom Lane wrote:
> Tomas Vondra <tv@fuzzy.cz> writes:
>> I believe the locale system (at the OS level) works just like before. I
>> remember I had to manually create the locales while initially setting up
>> the animals. Then, ~2 months ago something happened (I asssume a yum
>> update) and some of the locales disappeared. But I have recreated them,
>> except for sk_SK.WIN-1250. But the tests fail because of cs_CZ.WIN-1250
>> which does exist.
> 
> I am betting that you recreated them differently from before.

Aaaaand you're probably right. Apparently, I recreated them like this:
 $ localedef -v -c -i cs_CZ -f WIN-1250 cs_CZ.WIN-1250

but the correct way seems to be this:
 $ localedef -v -c -i cs_CZ -f CP1250 cs_CZ.WIN-1250

>> However when I tried to initialize a
>> cluster with cs_CZ.WIN-1250, I got an error like this:
> 
>> [pgbuild@regular-builds ~]$ pg_ctl -D tmp-data init
>> The files belonging to this database system will be owned by user "pgbuild".
>> This user must also own the server process.
> 
>> The database cluster will be initialized with locale "cs_CZ.WIN-1250".
>> could not determine encoding for locale "cs_CZ.WIN-1250": codeset is
>> "ANSI_X3.4-1968"
>> initdb: could not find suitable encoding for locale "cs_CZ.WIN-1250"
> 
> Locale cs_CZ.WIN-1250 is evidently marked with a codeset property of
> "ANSI_X3.4-1968" (which means old-school US-ASCII).  That's certainly
> wrong.  I believe the correct thing would be "CP1250".

Yes. I fixed the locales and added the locales back to the client
configuration.

regards
Tomas Vondra

Tomas Vondra <tv@fuzzy.cz> writes:
> On 20.12.2014 19:05, Tom Lane wrote:
>> I am betting that you recreated them differently from before.

> Aaaaand you're probably right. Apparently, I recreated them like this:

>   $ localedef -v -c -i cs_CZ -f WIN-1250 cs_CZ.WIN-1250

> but the correct way seems to be this:

>   $ localedef -v -c -i cs_CZ -f CP1250 cs_CZ.WIN-1250

Interesting.  Apparently, instead of failing outright on an unrecognized
charmap name, localedef just substituted ASCII and plowed ahead.  Bad dog.
        regards, tom lane

On 20.12.2014 07:39, Noah Misch wrote:<br /><span style="white-space: pre;">> Buildfarm members magpie, treepie and
fulmarwent absent on<br /> > 2014-10-29. Since returning on 2014-11-16, they have consistently<br /> > failed
with'initdb: invalid locale name "cs_CZ.WIN-1250"'. No<br /> > commits in that period readily explain a regression
inthis area. Did<br /> > the underlying system configurations change?<br /></span><br /> I'm pretty sure it's
becauseof broken locales at the system level. It was<br /> working fine, and I haven't done any substantial changes to
thesystem<br /> except for occasional "yum update", so I'm unaware of what went wong :(<br /><br /> The issue looks
likethis:<br /><br /> # locale -a | grep en_US<br /> en_US<br /> en_US.iso88591<br /> en_US.iso885915<br />
en_US.utf8<br/><br /> # locale en_US<br /> locale: unknown name "en_US"<br /><br /> The only reasons I can think of is
thatsome of the updates required<br /> a reboot, and I haven't done that because that would kill all the VMs<br />
runningon that HW, including the one with CLOBBER_CACHE_RECURSIVE <br /> tests. And that'd throw away tests running for
~3months.<br /><br /> I've disabled the three animals (magpie, fulmar, treepie) for now, because<br /> there's no point
inrunning the tests until the locale issues are fixed. If<br /> anyone has an idea of what might be wrong, let me
know.<br/><br /> Tomas<br /> 

On 20.12.2014 19:35, Tom Lane wrote:
> Tomas Vondra <tv@fuzzy.cz> writes:
>> On 20.12.2014 19:05, Tom Lane wrote:
>>> I am betting that you recreated them differently from before.
> 
>> Aaaaand you're probably right. Apparently, I recreated them like this:
> 
>>   $ localedef -v -c -i cs_CZ -f WIN-1250 cs_CZ.WIN-1250
> 
>> but the correct way seems to be this:
> 
>>   $ localedef -v -c -i cs_CZ -f CP1250 cs_CZ.WIN-1250
> 
> Interesting.  Apparently, instead of failing outright on an unrecognized
> charmap name, localedef just substituted ASCII and plowed ahead.  Bad dog.

Not really. It's rather about abusive owner of the dog, using '-c' to
force the dog to create the locale even when there are warings:
 # localedef -i cs_CZ -f WIN-1250 cs_CZ.WIN-1250 character map file `WIN-1250' not found: No such file or directory no
outputfile produced because warnings were issued
 

In my defense, I've been using verbose mode, and that produces a lot of
warnings like 'non-symbolic character value should not be used' (which
gets ignored in non-verbose mode) and thus missed the one important one.

regards
Tomas

On Sat, Dec 20, 2014 at 07:28:33PM +0100, Tomas Vondra wrote:
> On 20.12.2014 19:05, Tom Lane wrote:
> > Locale cs_CZ.WIN-1250 is evidently marked with a codeset property of
> > "ANSI_X3.4-1968" (which means old-school US-ASCII).  That's certainly
> > wrong.  I believe the correct thing would be "CP1250".
> 
> Yes. I fixed the locales and added the locales back to the client
> configuration.

Thanks.  These animals are now OK except on REL9_0_STABLE.  The log message of
commit 2dfa15d explains their ongoing 9.0 failures.  I recommend adding
--skip-steps=pl-install-check to their 9.0 invocations.  Dropping the affected
locales is another option, but we benefit from the rare encoding coverage more
than we benefit from the src/bin/pl coverage.

On 23.12.2014 09:19, Noah Misch wrote:
> On Sat, Dec 20, 2014 at 07:28:33PM +0100, Tomas Vondra wrote:
>> On 20.12.2014 19:05, Tom Lane wrote:
>>> Locale cs_CZ.WIN-1250 is evidently marked with a codeset property of
>>> "ANSI_X3.4-1968" (which means old-school US-ASCII).  That's certainly
>>> wrong.  I believe the correct thing would be "CP1250".
>>
>> Yes. I fixed the locales and added the locales back to the client
>> configuration.
> 
> Thanks.  These animals are now OK except on REL9_0_STABLE.  The log message of
> commit 2dfa15d explains their ongoing 9.0 failures.  I recommend adding
> --skip-steps=pl-install-check to their 9.0 invocations.  Dropping the affected
> locales is another option, but we benefit from the rare encoding coverage more
> than we benefit from the src/bin/pl coverage.

OK, can do. Something like this in build-farm.conf should work, right?

if ($branch eq 'REL9_0_STABLE')
{   push(@{$conf{config_opts}},"--skip-steps=pl-install-check");
}


regards
Tomas

On 12/23/2014 07:14 AM, Tomas Vondra wrote:
> On 23.12.2014 09:19, Noah Misch wrote:
>> On Sat, Dec 20, 2014 at 07:28:33PM +0100, Tomas Vondra wrote:
>>> On 20.12.2014 19:05, Tom Lane wrote:
>>>> Locale cs_CZ.WIN-1250 is evidently marked with a codeset property of
>>>> "ANSI_X3.4-1968" (which means old-school US-ASCII).  That's certainly
>>>> wrong.  I believe the correct thing would be "CP1250".
>>> Yes. I fixed the locales and added the locales back to the client
>>> configuration.
>> Thanks.  These animals are now OK except on REL9_0_STABLE.  The log message of
>> commit 2dfa15d explains their ongoing 9.0 failures.  I recommend adding
>> --skip-steps=pl-install-check to their 9.0 invocations.  Dropping the affected
>> locales is another option, but we benefit from the rare encoding coverage more
>> than we benefit from the src/bin/pl coverage.
> OK, can do. Something like this in build-farm.conf should work, right?
>
> if ($branch eq 'REL9_0_STABLE')
> {
>      push(@{$conf{config_opts}},"--skip-steps=pl-install-check");
> }
>


No, config_opts is what's passed to configure. Try something like:
    if ($branch eq 'REL9_0_STABLE')    {        $skip_steps{'pl-install-check'} = 1;    }

cheers

andrew

On 23.12.2014 15:21, Andrew Dunstan wrote:
> 
> No, config_opts is what's passed to configure. Try something like:
> 
>     if ($branch eq 'REL9_0_STABLE')
>     {
>         $skip_steps{'pl-install-check'} = 1;
>     }

Applied to all three animals.

Tomas

On Tue, Dec 23, 2014 at 03:32:59PM +0100, Tomas Vondra wrote:
> On 23.12.2014 15:21, Andrew Dunstan wrote:
> > 
> > No, config_opts is what's passed to configure. Try something like:
> > 
> >     if ($branch eq 'REL9_0_STABLE')
> >     {
> >         $skip_steps{'pl-install-check'} = 1;
> >     }
> 
> Applied to all three animals.

As of the time of this change, the animals ceased to report in.

On 28.12.2014 00:46, Noah Misch wrote:
> On Tue, Dec 23, 2014 at 03:32:59PM +0100, Tomas Vondra wrote:
>> On 23.12.2014 15:21, Andrew Dunstan wrote:
>>>
>>> No, config_opts is what's passed to configure. Try something like:
>>>
>>>     if ($branch eq 'REL9_0_STABLE')
>>>     {
>>>         $skip_steps{'pl-install-check'} = 1;
>>>     }
>>
>> Applied to all three animals.
> 
> As of the time of this change, the animals ceased to report in.

The strange thing is that while the first run worked fine (as
illustrated by the results submitted to pgbuildfarm.org), all the
following runs fail like this:

Global symbol "%skip_steps" requires explicit package name at
build-farm.conf line 308.
Compilation failed in require at ./run_branches.pl line 55.

Apparently, something is broken, but my Perl-fu is limited so I have no
idea what/why :-(

regards
Tomas

On 12/31/2014 07:49 AM, Tomas Vondra wrote:
> On 28.12.2014 00:46, Noah Misch wrote:
>> On Tue, Dec 23, 2014 at 03:32:59PM +0100, Tomas Vondra wrote:
>>> On 23.12.2014 15:21, Andrew Dunstan wrote:
>>>> No, config_opts is what's passed to configure. Try something like:
>>>>
>>>>      if ($branch eq 'REL9_0_STABLE')
>>>>      {
>>>>          $skip_steps{'pl-install-check'} = 1;
>>>>      }
>>> Applied to all three animals.
>> As of the time of this change, the animals ceased to report in.
> The strange thing is that while the first run worked fine (as
> illustrated by the results submitted to pgbuildfarm.org), all the
> following runs fail like this:
>
> Global symbol "%skip_steps" requires explicit package name at
> build-farm.conf line 308.
> Compilation failed in require at ./run_branches.pl line 55.
>
> Apparently, something is broken, but my Perl-fu is limited so I have no
> idea what/why :-(
>



Sorry, I should have tested it. This seems to work:
   if ($branch eq 'REL9_0_STABLE')   {        $PGBuild::Options::skip_steps .= ' pl-install-check';
$main::skip_steps{'pl-install-check'}= 1;   }
 

cheers

andrew

On 31.12.2014 17:29, Andrew Dunstan wrote:
> 
> Sorry, I should have tested it. This seems to work:
> 
>    if ($branch eq 'REL9_0_STABLE')
>    {
>         $PGBuild::Options::skip_steps .= ' pl-install-check';
>         $main::skip_steps{'pl-install-check'} = 1;
>    }
> 
> cheers

Meh, no problem. We've fixed it in 2014, so it's OK. To quote one of my
colleagues "I haven't tested it, I believe it works correctly." ;-)

Any ideas why it worked for the first run?

regards
Tomas

On 12/31/2014 12:26 PM, Tomas Vondra wrote:
> On 31.12.2014 17:29, Andrew Dunstan wrote:
>> Sorry, I should have tested it. This seems to work:
>>
>>     if ($branch eq 'REL9_0_STABLE')
>>     {
>>          $PGBuild::Options::skip_steps .= ' pl-install-check';
>>          $main::skip_steps{'pl-install-check'} = 1;
>>     }
>>
>> cheers
> Meh, no problem. We've fixed it in 2014, so it's OK. To quote one of my
> colleagues "I haven't tested it, I believe it works correctly." ;-)
>
> Any ideas why it worked for the first run?
>

No. It failed for me right off the bat.

cheers

andrew