Re: [BuildBot] News and announcements

Re: [BuildBot] News and announcements

[David Edelsohn]

On Wed, Jun 17, 2015 at 5:10PM, Doug Evans wrote:
> On Wed, Jun 17, 2015 at 1:09 PM, Sergio Durigan Junior
<sergiodj@sergiodj.net> wrote:

>> I also tried to dedicate some time to put a buildslave on an AIX machine
>> (in the GCC Farm).  This was a request by David Edelsohn.  However, I
>> did not manage to compile GDB there, and after spending some time trying
>> to figure out why, I had to switch context to another task.  If you are
>> able to help with this, I would really appreciate.
>
> I have successfully run the gdb testsuite on the aix machine on the gcc farm,
> so it is doable. I can (try to) help (if I can find the time).

Doug and I both have been able to build GDB and run the testsuite on
the system.  Is this a sign that current GDB has regressed on AIX?

- David

Re: [BuildBot] News and announcements

[David Edelsohn]

On Wed, Jun 17, 2015 at 5:10PM, Doug Evans wrote:
> On Wed, Jun 17, 2015 at 1:09 PM, Sergio Durigan Junior
<sergiodj@sergiodj.net> wrote:

>> I also tried to dedicate some time to put a buildslave on an AIX machine
>> (in the GCC Farm).  This was a request by David Edelsohn.  However, I
>> did not manage to compile GDB there, and after spending some time trying
>> to figure out why, I had to switch context to another task.  If you are
>> able to help with this, I would really appreciate.
>
> I have successfully run the gdb testsuite on the aix machine on the gcc farm,
> so it is doable. I can (try to) help (if I can find the time).

Sergio,

I currently am running a buildslave for CPython on the AIX system and
Olly Betts is running a buildslave for another project.

I built the latest snapshot of GDB last night and ran the testsuite.
There were numerous errors, but it ran.

I installed Buildslave 0.8.12 today.

I do not understand why you are unable to configure and build GDB on the system.

Thanks, David

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Sunday, June 21 2015, David Edelsohn wrote:

> Sergio,
>
> I currently am running a buildslave for CPython on the AIX system and
> Olly Betts is running a buildslave for another project.
>
> I built the latest snapshot of GDB last night and ran the testsuite.
> There were numerous errors, but it ran.
>
> I installed Buildslave 0.8.12 today.
>
> I do not understand why you are unable to configure and build GDB on the system.

Hey David,

I decided to give it a try today, and managed to compile GDB on gcc111
after git pull'ing.  It seems like the tree was broken for AIX when I
was attempting to compile GDB there, some weeks ago.  One more reason to
get our BuildBot there!

For the record, this is what I did there:

# export PATH=/opt/freeware/bin:$PATH
# ../binutils-gdb/configure \
      --disable-{binutils,gprof,ld,gold,gas,werror}
# gmake -j4 MAKEINFO=true

Yeah, not neat...

I will set the GDB buildslave there in a few moments, and then I'll let
you know.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Monday, June 22 2015, I wrote:

> I will set the GDB buildslave there in a few moments, and then I'll let
> you know.

Everything is set up.  Now let's wait until there's a commit to be
built, and check if everything is working fine.

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Monday, June 22 2015, I wrote:

> On Monday, June 22 2015, I wrote:
>
>> I will set the GDB buildslave there in a few moments, and then I'll let
>> you know.
>
> Everything is set up.  Now let's wait until there's a commit to be
> built, and check if everything is working fine.

After dealing with some minor issues, I'm now seeing this:

  <http://gdb-build.sergiodj.net/builders/AIX-POWER7-plain/builds/2/steps/compile%20gdb/logs/stdio>

I talked to Bergner, who checked this patch in, and he said he doesn't
know why this is happing.  I'm also not really interested (and most
importantly, don't have the time) in maintaining the AIX builder(s), so
if you could take a look at this I'd appreciate.

As a side note, and since you are the interested part in this AIX
buildslave, I was wondering if you want to be the responsible for it.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[David Edelsohn]

On Mon, Jun 22, 2015 at 9:18 PM, Sergio Durigan Junior
<sergiodj@redhat.com> wrote:
> On Monday, June 22 2015, I wrote:
>
>> On Monday, June 22 2015, I wrote:
>>
>>> I will set the GDB buildslave there in a few moments, and then I'll let
>>> you know.
>>
>> Everything is set up.  Now let's wait until there's a commit to be
>> built, and check if everything is working fine.
>
> After dealing with some minor issues, I'm now seeing this:
>
>   <http://gdb-build.sergiodj.net/builders/AIX-POWER7-plain/builds/2/steps/compile%20gdb/logs/stdio>
>
> I talked to Bergner, who checked this patch in, and he said he doesn't
> know why this is happing.  I'm also not really interested (and most
> importantly, don't have the time) in maintaining the AIX builder(s), so
> if you could take a look at this I'd appreciate.
>
> As a side note, and since you are the interested part in this AIX
> buildslave, I was wondering if you want to be the responsible for it.

I restarted the buildslave with /opt/freeware/bin in your path, as we
had discussed earlier.

We saw a similar error building GCC a few months ago after the CHKP
patches.  This means that GDB now has a debug statement that is more
than 64K in length.  More recent versions of GCC allow debug
continuation lines, but this really means that m32c has a ridiculously
large enum.  It's also bloating the rest of the GDB object files, even
if DWARF can handle the length.

This error means that m32c has a real problem and AIX simply was the
canary.  Debug information that large is allowed but is evidence of a
design problem.

Yes, I can run the buildslave, if you want.

 - David

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Monday, June 22 2015, David Edelsohn wrote:

>> After dealing with some minor issues, I'm now seeing this:
>>
>>   <http://gdb-build.sergiodj.net/builders/AIX-POWER7-plain/builds/2/steps/compile%20gdb/logs/stdio>
>>
>> I talked to Bergner, who checked this patch in, and he said he doesn't
>> know why this is happing.  I'm also not really interested (and most
>> importantly, don't have the time) in maintaining the AIX builder(s), so
>> if you could take a look at this I'd appreciate.
>>
>> As a side note, and since you are the interested part in this AIX
>> buildslave, I was wondering if you want to be the responsible for it.
>
> I restarted the buildslave with /opt/freeware/bin in your path, as we
> had discussed earlier.

As you can see in the link above, the buildslave already has
/opt/freeware/bin in the PATH, therefore everything was already covered
in this area before.  But thanks for this.

> We saw a similar error building GCC a few months ago after the CHKP
> patches.  This means that GDB now has a debug statement that is more
> than 64K in length.  More recent versions of GCC allow debug
> continuation lines, but this really means that m32c has a ridiculously
> large enum.  It's also bloating the rest of the GDB object files, even
> if DWARF can handle the length.

Right.  Strangely I wasn't seeing this problem when compiling GDB
manually in the machine, and Peter's commit doesn't touch the m32c file.

> Yes, I can run the buildslave, if you want.

That would be awesome!  It is currently running in my homedir at gcc111,
but I can stop it so that you can configure the buildslave at your
homedir if you want.  I'd prefer to keep the same name (aix-power7-1).
I can also send you (in private) the password for the buildslave, along
with the full command to create it.  Let me know when/what you need and
I'll be happy to help.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[David Edelsohn]

On Mon, Jun 22, 2015 at 9:44 PM, Sergio Durigan Junior
<sergiodj@redhat.com> wrote:
> On Monday, June 22 2015, David Edelsohn wrote:
>
>>> After dealing with some minor issues, I'm now seeing this:
>>>
>>>   <http://gdb-build.sergiodj.net/builders/AIX-POWER7-plain/builds/2/steps/compile%20gdb/logs/stdio>
>>>
>>> I talked to Bergner, who checked this patch in, and he said he doesn't
>>> know why this is happing.  I'm also not really interested (and most
>>> importantly, don't have the time) in maintaining the AIX builder(s), so
>>> if you could take a look at this I'd appreciate.
>>>
>>> As a side note, and since you are the interested part in this AIX
>>> buildslave, I was wondering if you want to be the responsible for it.
>>
>> I restarted the buildslave with /opt/freeware/bin in your path, as we
>> had discussed earlier.
>
> As you can see in the link above, the buildslave already has
> /opt/freeware/bin in the PATH, therefore everything was already covered
> in this area before.  But thanks for this.

If you have /opt/freeware/bin in your path, you do not need to use gmake.

>> We saw a similar error building GCC a few months ago after the CHKP
>> patches.  This means that GDB now has a debug statement that is more
>> than 64K in length.  More recent versions of GCC allow debug
>> continuation lines, but this really means that m32c has a ridiculously
>> large enum.  It's also bloating the rest of the GDB object files, even
>> if DWARF can handle the length.
>
> Right.  Strangely I wasn't seeing this problem when compiling GDB
> manually in the machine, and Peter's commit doesn't touch the m32c file.

Something increased the size of an enum (likely) or other single
debugging entry larger than 64K.

>
>> Yes, I can run the buildslave, if you want.
>
> That would be awesome!  It is currently running in my homedir at gcc111,
> but I can stop it so that you can configure the buildslave at your
> homedir if you want.  I'd prefer to keep the same name (aix-power7-1).
> I can also send you (in private) the password for the buildslave, along
> with the full command to create it.  Let me know when/what you need and
> I'll be happy to help.

I will transfer the info.

- David

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Monday, June 22 2015, David Edelsohn wrote:

>>> I restarted the buildslave with /opt/freeware/bin in your path, as we
>>> had discussed earlier.
>>
>> As you can see in the link above, the buildslave already has
>> /opt/freeware/bin in the PATH, therefore everything was already covered
>> in this area before.  But thanks for this.
>
> If you have /opt/freeware/bin in your path, you do not need to use gmake.

Cool, didn't know that.

>>> Yes, I can run the buildslave, if you want.
>>
>> That would be awesome!  It is currently running in my homedir at gcc111,
>> but I can stop it so that you can configure the buildslave at your
>> homedir if you want.  I'd prefer to keep the same name (aix-power7-1).
>> I can also send you (in private) the password for the buildslave, along
>> with the full command to create it.  Let me know when/what you need and
>> I'll be happy to help.
>
> I will transfer the info.

Thanks.  I updated the contact address internally, so that if the
buildslave suddenly stops you'll get notified.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[David Edelsohn]

On Mon, Jun 22, 2015 at 10:11 PM, Sergio Durigan Junior
<sergiodj@redhat.com> wrote:
> On Monday, June 22 2015, David Edelsohn wrote:
>
>>>> I restarted the buildslave with /opt/freeware/bin in your path, as we
>>>> had discussed earlier.
>>>
>>> As you can see in the link above, the buildslave already has
>>> /opt/freeware/bin in the PATH, therefore everything was already covered
>>> in this area before.  But thanks for this.
>>
>> If you have /opt/freeware/bin in your path, you do not need to use gmake.
>
> Cool, didn't know that.
>
>>>> Yes, I can run the buildslave, if you want.
>>>
>>> That would be awesome!  It is currently running in my homedir at gcc111,
>>> but I can stop it so that you can configure the buildslave at your
>>> homedir if you want.  I'd prefer to keep the same name (aix-power7-1).
>>> I can also send you (in private) the password for the buildslave, along
>>> with the full command to create it.  Let me know when/what you need and
>>> I'll be happy to help.
>>
>> I will transfer the info.
>
> Thanks.  I updated the contact address internally, so that if the
> buildslave suddenly stops you'll get notified.

I just checked out binutils-gdb and built it manually without problem.

Can you set the build on AIX to not use "--enable-targets=all"?

Thanks, David

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Monday, June 22 2015, David Edelsohn wrote:

> I just checked out binutils-gdb and built it manually without problem.
>
> Can you set the build on AIX to not use "--enable-targets=all"?

Done.

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[David Edelsohn]

On Mon, Jun 22, 2015 at 10:25 PM, Sergio Durigan Junior
<sergiodj@redhat.com> wrote:
> On Monday, June 22 2015, David Edelsohn wrote:
>
>> I just checked out binutils-gdb and built it manually without problem.
>>
>> Can you set the build on AIX to not use "--enable-targets=all"?
>
> Done.

And "success".

- David

Re: [BuildBot] News and announcements

[David Edelsohn]

Sergio,

If you configure another builder for Debian Wheezy on Linux for System
z and send me the private host information, I can set up a GDB
buildbot for Linux on System z.

Thanks, David

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Tuesday, June 23 2015, David Edelsohn wrote:

> Sergio,
>
> If you configure another builder for Debian Wheezy on Linux for System
> z and send me the private host information, I can set up a GDB
> buildbot for Linux on System z.

Sure.  I need the following information:

- How many jobs can the buildslave handle?  By that I basically mean
  "what is N in make -j<N>"?

- I will name the buildslave as "debian-wheezy-s390x-1", is that OK for
  you?

- I will assign 3 builders to the buildslave: Debian-s390x-m64,
  Debian-s390x-native-gdbserver-m64 and
  Debian-s390x-native-extended-gdbserver-m64.  Are those OK for you?

Once I have all the information needed, I will send the password to you
in private.  Also, there are instructions on how to prepare the
buildslave for GDB here:

  <https://sourceware.org/gdb/wiki/BuildBot>

Including packages that could be installed to increase the coverage of
the testsuite, etc.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Friday, June 19 2015, Yao Qi wrote:

> Sergio Durigan Junior <sergiodj@sergiodj.net> writes:
>
>> Needless to say, we also welcome donations of new machines for our
>> BuildBot!  For example, we don't have an ARM buildslave yet, which would
>> be very nice.  So if you're interested in helping us, you can either
>> take a look at this wiki page:
>>
>>   <https://sourceware.org/gdb/wiki/BuildBot>
>
> Hi Sergio,
> Does buildmaster need to access to buildslave?  Is it OK that the
> buildslave is put behind the firewall?  I'd like to know how these
> arm or aarch64 buildslaves are used in BuildBot, and see how to provide
> them from linaro or arm (and whether it is allowed to do so).

Hey Yao,

The buildslave is responsible for starting the connection, so the
buildmaster does not need to have direct access to it.  I myself run a
buildslave behind a firewall/NAT, and there is no problem.  All you need
to do is to set up the buildslave (as described in the wiki), tell me
about it (so that I can include it on buildmaster and generate a
password for you), and then start it.

The buildmaster will issue build requests to the buildslave, which will
build, run the tests, and the report back to buildmaster.  That's it.
Nobody needs to have access to the buildslaves.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Wednesday, June 17 2015, Doug Evans wrote:

> I have successfully run the gdb testsuite on the aix machine on the gcc farm,
> so it is doable. I can (try to) help (if I can find the time).

Thanks, Doug.  I will let you and Joel know when I have time to tackle
this again.

> Let me take the opportunity repeat some of my feature requests. :-)
>
> Instead of running a build per commit, just pick up the current tree
> on each iteration of the builder (and early exit if nothing has changed).

I don't know if I understood the request well, but there is no such
thing as "iteration of the builder".  The way BuildBot works is: if
there is a new commit in the repository, it triggers builds for every
builder (a.k.a. "build configurations").  This makes each builder to
update its own copy of the upstream repository, and build the specified
commit (usually HEAD, but it's possible to force them to build other
commits as well).

> And if there's a problem, bisect.

With the way BuildBot works (described above), there is no need to
bisect something: if a commit introduces a (valid) failure on the
testsuite, we will know right away (i.e., when the builders build it).
Also, and as I said in my first message, if the commit breaks GDB's
compilation, we will also know right away, and the commit author will be
notified directly.

Hm, but now I think I understood what you proposed above.  You are
proposing to make something like nightly builders, is that right?  I.e.,
instead of building every commit in the repository, just build git HEAD
at the same time every day.  That can be done with BuildBot (although I
don't know how hard it would be to implement a "bisect").

> This should scale better when there are more frequent commits.

I agree, although I like the idea of testing every commit as it hits the
tree.

> Commits usually don't break the build so picking up the tree
> where it is should be just fine (has been for me with my
> nightly build of trunk for years).
>
> If a test fails, run it several times.
> This should help catch hard fails vs flaky fails.

Yeah, this last part would be neat (although I am also not sure how hard
it would be to implement this on BuildBot).

> IWBN if one build handled multiple test configs.

This was something I could have thought about when I was designing how
our BuildBot was going to work.  Currently, we have one builder for
every build & test configuration.  I agree it should be possible to have
e.g. one builder to test multiple test configs...  This will require a
major redesign of how our BuildBot works, so I don't know when I'll have
the time to tackle this.

> As for new test configs I'd like to see:
> gcc vs llvm
> gdb-generated index vs gold-generated index vs no index
> dwarf2 vs dwarf4 (with type units)
> fission vs fission-with-dwp vs no-fission
> stabs
> gdbserver vs no gdbserver
> Each of these would use the same gdb, it's just
> the test run that's different.

Noted.  I don't know if it will be possible to easily implement the "vs"
configs; what can be done is running "gcc" on a builder and then "llvm"
on another...

>> Problematic testcases
>> =====================
>>
>> Efforts on reducing the number of problematic testcases are being done,
>> mostly by Pedro.  Thanks, Pedro!  Hopefully, sometime in the future we
>> will reach a state when all the notifications of regressions will
>> actually be trustworthy.
>
> It's difficult to prevent flaky tests from being added in the future,
> which is why I'd like to see the build-bot help find them.

BuildBot is already "finding" them: you can compare the test results (on
gdb-testers) and see that there are lots of racy testcases being
reported for various builders.  As I may have mentioned before, there is
a way to make our BuildBot ignore tests that are racy, but this is just
a workaround, of course.

>>   - Update the existing x86_64 machine to Fedora 22
>>
>>   - Get the new x86_64 machines and set them up for our BuildBot
>>
>>   - Get the AIX buildslave up and running
>>
>>   - Continue trying to convince others to donate buildslaves to us :-)
>
> Working on it ...

Thanks.

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[Yao Qi]

Sergio Durigan Junior <sergiodj@sergiodj.net> writes:

> Needless to say, we also welcome donations of new machines for our
> BuildBot!  For example, we don't have an ARM buildslave yet, which would
> be very nice.  So if you're interested in helping us, you can either
> take a look at this wiki page:
>
>   <https://sourceware.org/gdb/wiki/BuildBot>

Hi Sergio,
Does buildmaster need to access to buildslave?  Is it OK that the
buildslave is put behind the firewall?  I'd like to know how these
arm or aarch64 buildslaves are used in BuildBot, and see how to provide
them from linaro or arm (and whether it is allowed to do so).

-- 
Yao (齐尧)

Re: [BuildBot] News and announcements

[Doug Evans]

On Wed, Jun 17, 2015 at 1:09 PM, Sergio Durigan Junior
<sergiodj@sergiodj.net> wrote:
> Hi there,
>
> It has been some time since I set our BuildBot instance at:
>
>   <http://gdb-build.sergiodj.net>
>
> And I figured I could send a message announcing some news here.
>
> New buildslaves
> ===============
>
> First of all, I would like to thank IBM again for the old POWER7
> buildslaves by new, more capable POWER8 VM's.  We now have one PPC64BE
> and one PPC64LE buildslaves, both running Fedora 22.  Special thanks to
> Tulio Magno Quites Machado Filho and to Edjunior Machado, both from IBM
> Brazil, for all the help with the new VM's.  Much appreciated!
>
> Now, we have the following buildslaves in our BuildBot:
>
>   - One Fedora 21 x86_64 machine (owned by me; more expected to come in
>     the next months)
>
>   - One Fedora 22 PPC64BE machine (owned by Edjunior Machado)
>
>   - One Fedora 22 PPC64LE machine (owned by Edjunior Machado)
>
>   - One Fedora 21 s390x machine (owned by Mark Wielaard)
>
>   - One Debian Wheezy x86_64 machine (owned by Mark Wielaard)
>
>   - One Debian Wheezy i686 machine (owned by Mark Wielaard)
>
> (There are also 4 FreeBSD buildslaves, but they are currently inactive.)
>
> As I said, I expect to receive two more x86_64 machines in the next
> months; I intend to install Fedora in at least one of them, because the
> current Fedora x86_64 machine (owned by me) is struggling to keep the
> pace with the other buildslaves.
>
> I also tried to dedicate some time to put a buildslave on an AIX machine
> (in the GCC Farm).  This was a request by David Edelsohn.  However, I
> did not manage to compile GDB there, and after spending some time trying
> to figure out why, I had to switch context to another task.  If you are
> able to help with this, I would really appreciate.

I have successfully run the gdb testsuite on the aix machine on the gcc farm,
so it is doable. I can (try to) help (if I can find the time).

>
> Needless to say, we also welcome donations of new machines for our
> BuildBot!  For example, we don't have an ARM buildslave yet, which would
> be very nice.  So if you're interested in helping us, you can either
> take a look at this wiki page:
>
>   <https://sourceware.org/gdb/wiki/BuildBot>
>
> Or contact me directly.  The more machines we have, the more we can test
> GDB :-).
>
> New notifications
> =================
>
> If you commit a piece of code that breaks GDB in any of the current
> builders, you will now receive a direct e-mail from BuildBot.  This was
> an old feature I had in mind, and Mark Wielaard put some pressure to
> implement it :-).  The way it is implemented today, you will receive a
> message for every builder that is not able to compile your code.  This
> means that you can end up receiving lots of e-mails (currently, we have
> 24 builders).  While I agree that this may be annoying, well, breaking
> GDB is also annoying, so consider this a punishment!  Also, it is not
> uncommon to see commits that break only specific builders, so I think it
> is important to trigger a new notification every time something bad
> happens.
>
> I have also enabled a new notification for when a buildslave goes
> offline.  I decided to do that after having problems with buildslaves
> that were unstable and required constant monitoring.  Now, BuildBot
> automatically sends an e-mail to the buildslave's owner, notifying him.
>
> New builders
> ============
>
> Pedro asked for a new builder to test just the compilation of the new
> --enable-build-with-cxx flag.  It is now created on:
>
>   <http://gdb-build.sergiodj.net/builders/Fedora-x86_64-cxx-build-m64>
>
> This builder just compiles GDB, i.e., it does not run the testsuite like
> the others.
>
> Doug asked me if it was possible to create a new builder to test things
> with --target_board=stabs.  Unfortunately, we don't have resources to
> create this builder on the x86_64 machines; I will try to create the
> builder on the PPC64 buildslaves and see how they behave.

Let me take the opportunity repeat some of my feature requests. :-)

Instead of running a build per commit, just pick up the current tree
on each iteration of the builder (and early exit if nothing has changed).
And if there's a problem, bisect.
This should scale better when there are more frequent commits.
Commits usually don't break the build so picking up the tree
where it is should be just fine (has been for me with my
nightly build of trunk for years).

If a test fails, run it several times.
This should help catch hard fails vs flaky fails.

IWBN if one build handled multiple test configs.

As for new test configs I'd like to see:
gcc vs llvm
gdb-generated index vs gold-generated index vs no index
dwarf2 vs dwarf4 (with type units)
fission vs fission-with-dwp vs no-fission
stabs
gdbserver vs no gdbserver
Each of these would use the same gdb, it's just
the test run that's different.

>
> Problematic testcases
> =====================
>
> Efforts on reducing the number of problematic testcases are being done,
> mostly by Pedro.  Thanks, Pedro!  Hopefully, sometime in the future we
> will reach a state when all the notifications of regressions will
> actually be trustworthy.

It's difficult to prevent flaky tests from being added in the future,
which is why I'd like to see the build-bot help find them.

>
> And that's it.  Just a remind, all the reports generated by the BuildBot
> are publicly accessible on the gdb-testers mailing list:
>
>   <https://sourceware.org/ml/gdb-testers/>
>
> This is now a very high-volume mailing list, so be aware if you decide
> to subscribe to it!
>
> Other than that, I am looking forward to improving the BuildBot in any
> way that the project needs (taking into consideration the amount of
> spare time I have, of course).  My next steps are:
>
>   - Try to fit the builder requested by Doug using the current resources

My internal nightly test runs cover 16 different test configs
(alluded to above).
They take a few hours to run with a 6 core machine.

>
>   - Update the existing x86_64 machine to Fedora 22
>
>   - Get the new x86_64 machines and set them up for our BuildBot
>
>   - Get the AIX buildslave up and running
>
>   - Continue trying to convince others to donate buildslaves to us :-)

Working on it ...

> Any help is appreciated, of course.
>
> Cheers,

Thanks for setting this up!

Re: [BuildBot] News and announcements

[Sergio Durigan Junior]

On Wednesday, June 17 2015, Simon Marchi wrote:

> Hi Sergio,

Hey Simon,

> One thing that I would like to see covered would be Python 2 vs Python
> 3. Is there a builder that builds gdb with Python 3? It's not very
> arch-dependent, so I think that you could build with Python 3
> (--with-python=/usr/bin/python3 for example) on a single builder and
> it would be covered.

Nice.  None of the builders are building with Python 3, to the extent of
my knowledge.  I could try to allocate one more PPC* builder for that,
although my real preference would be to wait a little more until I have
the new x86_64 machines available.  I'll see what I can do :-).

> Thanks for this whole effort, it's _very_ valuable.

Thanks for contributing with ideas :-).

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [BuildBot] News and announcements

[Simon Marchi]

On 17 June 2015 at 14:09, Sergio Durigan Junior <sergiodj@sergiodj.net> wrote:
> Hi there,
>
> It has been some time since I set our BuildBot instance at:
>
>   <http://gdb-build.sergiodj.net>
>
> And I figured I could send a message announcing some news here.
>
> New buildslaves
> ===============
>
> First of all, I would like to thank IBM again for the old POWER7
> buildslaves by new, more capable POWER8 VM's.  We now have one PPC64BE
> and one PPC64LE buildslaves, both running Fedora 22.  Special thanks to
> Tulio Magno Quites Machado Filho and to Edjunior Machado, both from IBM
> Brazil, for all the help with the new VM's.  Much appreciated!
>
> Now, we have the following buildslaves in our BuildBot:
>
>   - One Fedora 21 x86_64 machine (owned by me; more expected to come in
>     the next months)
>
>   - One Fedora 22 PPC64BE machine (owned by Edjunior Machado)
>
>   - One Fedora 22 PPC64LE machine (owned by Edjunior Machado)
>
>   - One Fedora 21 s390x machine (owned by Mark Wielaard)
>
>   - One Debian Wheezy x86_64 machine (owned by Mark Wielaard)
>
>   - One Debian Wheezy i686 machine (owned by Mark Wielaard)
>
> (There are also 4 FreeBSD buildslaves, but they are currently inactive.)
>
> As I said, I expect to receive two more x86_64 machines in the next
> months; I intend to install Fedora in at least one of them, because the
> current Fedora x86_64 machine (owned by me) is struggling to keep the
> pace with the other buildslaves.
>
> I also tried to dedicate some time to put a buildslave on an AIX machine
> (in the GCC Farm).  This was a request by David Edelsohn.  However, I
> did not manage to compile GDB there, and after spending some time trying
> to figure out why, I had to switch context to another task.  If you are
> able to help with this, I would really appreciate.
>
> Needless to say, we also welcome donations of new machines for our
> BuildBot!  For example, we don't have an ARM buildslave yet, which would
> be very nice.  So if you're interested in helping us, you can either
> take a look at this wiki page:
>
>   <https://sourceware.org/gdb/wiki/BuildBot>
>
> Or contact me directly.  The more machines we have, the more we can test
> GDB :-).
>
> New notifications
> =================
>
> If you commit a piece of code that breaks GDB in any of the current
> builders, you will now receive a direct e-mail from BuildBot.  This was
> an old feature I had in mind, and Mark Wielaard put some pressure to
> implement it :-).  The way it is implemented today, you will receive a
> message for every builder that is not able to compile your code.  This
> means that you can end up receiving lots of e-mails (currently, we have
> 24 builders).  While I agree that this may be annoying, well, breaking
> GDB is also annoying, so consider this a punishment!  Also, it is not
> uncommon to see commits that break only specific builders, so I think it
> is important to trigger a new notification every time something bad
> happens.
>
> I have also enabled a new notification for when a buildslave goes
> offline.  I decided to do that after having problems with buildslaves
> that were unstable and required constant monitoring.  Now, BuildBot
> automatically sends an e-mail to the buildslave's owner, notifying him.
>
> New builders
> ============
>
> Pedro asked for a new builder to test just the compilation of the new
> --enable-build-with-cxx flag.  It is now created on:
>
>   <http://gdb-build.sergiodj.net/builders/Fedora-x86_64-cxx-build-m64>
>
> This builder just compiles GDB, i.e., it does not run the testsuite like
> the others.
>
> Doug asked me if it was possible to create a new builder to test things
> with --target_board=stabs.  Unfortunately, we don't have resources to
> create this builder on the x86_64 machines; I will try to create the
> builder on the PPC64 buildslaves and see how they behave.
>
> Problematic testcases
> =====================
>
> Efforts on reducing the number of problematic testcases are being done,
> mostly by Pedro.  Thanks, Pedro!  Hopefully, sometime in the future we
> will reach a state when all the notifications of regressions will
> actually be trustworthy.
>
>
>
> And that's it.  Just a remind, all the reports generated by the BuildBot
> are publicly accessible on the gdb-testers mailing list:
>
>   <https://sourceware.org/ml/gdb-testers/>
>
> This is now a very high-volume mailing list, so be aware if you decide
> to subscribe to it!
>
> Other than that, I am looking forward to improving the BuildBot in any
> way that the project needs (taking into consideration the amount of
> spare time I have, of course).  My next steps are:
>
>   - Try to fit the builder requested by Doug using the current resources
>
>   - Update the existing x86_64 machine to Fedora 22
>
>   - Get the new x86_64 machines and set them up for our BuildBot
>
>   - Get the AIX buildslave up and running
>
>   - Continue trying to convince others to donate buildslaves to us :-)
>
> Any help is appreciated, of course.
>
> Cheers,
>
> --
> Sergio
> GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
> Please send encrypted e-mail if possible
> http://sergiodj.net/

Hi Sergio,

One thing that I would like to see covered would be Python 2 vs Python
3. Is there a builder that builds gdb with Python 3? It's not very
arch-dependent, so I think that you could build with Python 3
(--with-python=/usr/bin/python3 for example) on a single builder and
it would be covered.

Thanks for this whole effort, it's _very_ valuable.

Simon

[BuildBot] News and announcements

[Sergio Durigan Junior]

Hi there,

It has been some time since I set our BuildBot instance at:

  <http://gdb-build.sergiodj.net>

And I figured I could send a message announcing some news here.

New buildslaves
===============

First of all, I would like to thank IBM again for the old POWER7
buildslaves by new, more capable POWER8 VM's.  We now have one PPC64BE
and one PPC64LE buildslaves, both running Fedora 22.  Special thanks to
Tulio Magno Quites Machado Filho and to Edjunior Machado, both from IBM
Brazil, for all the help with the new VM's.  Much appreciated!

Now, we have the following buildslaves in our BuildBot:

  - One Fedora 21 x86_64 machine (owned by me; more expected to come in
    the next months)

  - One Fedora 22 PPC64BE machine (owned by Edjunior Machado)

  - One Fedora 22 PPC64LE machine (owned by Edjunior Machado)

  - One Fedora 21 s390x machine (owned by Mark Wielaard)

  - One Debian Wheezy x86_64 machine (owned by Mark Wielaard)

  - One Debian Wheezy i686 machine (owned by Mark Wielaard)

(There are also 4 FreeBSD buildslaves, but they are currently inactive.)

As I said, I expect to receive two more x86_64 machines in the next
months; I intend to install Fedora in at least one of them, because the
current Fedora x86_64 machine (owned by me) is struggling to keep the
pace with the other buildslaves.

I also tried to dedicate some time to put a buildslave on an AIX machine
(in the GCC Farm).  This was a request by David Edelsohn.  However, I
did not manage to compile GDB there, and after spending some time trying
to figure out why, I had to switch context to another task.  If you are
able to help with this, I would really appreciate.

Needless to say, we also welcome donations of new machines for our
BuildBot!  For example, we don't have an ARM buildslave yet, which would
be very nice.  So if you're interested in helping us, you can either
take a look at this wiki page:

  <https://sourceware.org/gdb/wiki/BuildBot>

Or contact me directly.  The more machines we have, the more we can test
GDB :-).

New notifications
=================

If you commit a piece of code that breaks GDB in any of the current
builders, you will now receive a direct e-mail from BuildBot.  This was
an old feature I had in mind, and Mark Wielaard put some pressure to
implement it :-).  The way it is implemented today, you will receive a
message for every builder that is not able to compile your code.  This
means that you can end up receiving lots of e-mails (currently, we have
24 builders).  While I agree that this may be annoying, well, breaking
GDB is also annoying, so consider this a punishment!  Also, it is not
uncommon to see commits that break only specific builders, so I think it
is important to trigger a new notification every time something bad
happens.

I have also enabled a new notification for when a buildslave goes
offline.  I decided to do that after having problems with buildslaves
that were unstable and required constant monitoring.  Now, BuildBot
automatically sends an e-mail to the buildslave's owner, notifying him.

New builders
============

Pedro asked for a new builder to test just the compilation of the new
--enable-build-with-cxx flag.  It is now created on:

  <http://gdb-build.sergiodj.net/builders/Fedora-x86_64-cxx-build-m64>

This builder just compiles GDB, i.e., it does not run the testsuite like
the others.

Doug asked me if it was possible to create a new builder to test things
with --target_board=stabs.  Unfortunately, we don't have resources to
create this builder on the x86_64 machines; I will try to create the
builder on the PPC64 buildslaves and see how they behave.

Problematic testcases
=====================

Efforts on reducing the number of problematic testcases are being done,
mostly by Pedro.  Thanks, Pedro!  Hopefully, sometime in the future we
will reach a state when all the notifications of regressions will
actually be trustworthy.



And that's it.  Just a remind, all the reports generated by the BuildBot
are publicly accessible on the gdb-testers mailing list:

  <https://sourceware.org/ml/gdb-testers/>

This is now a very high-volume mailing list, so be aware if you decide
to subscribe to it!

Other than that, I am looking forward to improving the BuildBot in any
way that the project needs (taking into consideration the amount of
spare time I have, of course).  My next steps are:

  - Try to fit the builder requested by Doug using the current resources

  - Update the existing x86_64 machine to Fedora 22

  - Get the new x86_64 machines and set them up for our BuildBot

  - Get the AIX buildslave up and running

  - Continue trying to convince others to donate buildslaves to us :-)

Any help is appreciated, of course.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/