Master Shell Command exist status does not affect overall build status

Hello,

One of my buildslaves performs a couple of additional MasterShellCommands (output post processing)
Sometimes these steps fail and in the single build view they are marked as red but the overall status of the
slave's build process remains green.
I would suggest to make it configurable since the master shell command is important for the buildslave
status too (at least for my purpose) and I therefore need to check always the waterfall to see if all single
steps were run successfully.

Buildbot 0.8.3

Thanks
Sam
--

-- 
Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

Re: Master Shell Command exist status does not affect overall build status

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

_______________________________________________
Buildbot-devel mailing list
Buildbot-devel@...
https://lists.sourceforge.net/lists/listinfo/buildbot-devel

Re: Master Shell Command exist status does not affect overall build status

Hello,
I tried,
no effect. Still everything green

Sam
-------- Original-Nachricht --------
> Datum: Sat, 1 Jan 2011 09:39:41 -0600
> Von: Amber Yust <ayust@...>
> An: ariasgore@...
> CC: buildbot-devel@...
> Betreff: Re: [Buildbot-devel] Master Shell Command exist status does not affect overall build status

> Have you tried setting flunkOnFailure=True in the MasterShellCommand
> arguments?
> 
> ~Amber
> 
> On Sat, Jan 1, 2011 at 9:33 AM, <ariasgore@...> wrote:
> 
> > Hello,
> >
> > One of my buildslaves performs a couple of additional
> MasterShellCommands
> > (output post processing)
> > Sometimes these steps fail and in the single build view they are marked
> as
> > red but the overall status of the slave's build process remains green.
> > I would suggest to make it configurable since the master shell command
> is
> > important for the buildslave status too (at least for my purpose) and I
> > therefore need to check always the waterfall to see if all single steps
> were
> > run successfully.
> >
> > Buildbot 0.8.3
> >
> > Thanks
> > Sam
> > --
> > Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
> > belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de
> >
> >
> >
> ------------------------------------------------------------------------------
> > Learn how Oracle Real Application Clusters (RAC) One Node allows
> customers
> > to consolidate database storage, standardize their database environment,
> > and,
> > should the need arise, upgrade to a full multi-node Oracle RAC database
> > without downtime or disruption
> > http://p.sf.net/sfu/oracle-sfdevnl
> > _______________________________________________
> > Buildbot-devel mailing list
> > Buildbot-devel@...
> > https://lists.sourceforge.net/lists/listinfo/buildbot-devel
> >

--

-- 
Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

[Bug?] FileUpload more verbose on error

Hello,
when I use FileUpload to store a file into a master's directory that does not exist it simply passes and does
not fail.
So invalid paths do not create any stdio or stderr AND they simply pass.

Btw and even if a FileUpload fails entirely it does not affect the Buildbot Status (and it has no flunkOnError)

Sam
--

-- 
Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

GitPoller fixes in 0.8.3p1

The GitPoller in 0.8.3 doesn't work.  There have been a few bugs now
(#1745 and #1742) and several folks seeing problems in #buildbot.

In accordance with http://trac.buildbot.net/wiki/SupportPolicy, I'm
going to release 0.8.3p1 with fixes to the GitPoller.  The current
'master' is unstable because of the sqlalchemy work going on there, so
I cannot do a fresh 0.8.4 release without rolling much of that back.

Before I pull the trigger on 0.8.3p1, I'd like to get feedback that
this is working for some of the folks who had trouble.

If you're using GitPoller with 0.8.3 and have been having trouble,
please try installing
  http://buildbot.net/tmp/buildbot-0.8.3-2-g15fbae9.tar.gz
and let me know if the situation improves for you?

I've tested this by throwing 1000 changes at it, and it does not run
out of file descriptors.  It also correctly initializes a fresh
repository on startup, and does not re-create it on subsequent
startups.

If there are other critical bugs with simple, shovel-ready fixes that
you'd like to see in 0.8.3p1, please let me know.  I don't want an
open season on every defect, but we have fixes available we might as
well ship them sooner rather than later.

Dustin

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

Re: Master Shell Command exist status does not affect overall build status

On Sat, Jan 1, 2011 at 12:09 PM,  <ariasgore@...> wrote:
> I tried,
> no effect. Still everything green

I had a look, and MasterShellCommand's flunkOnFailure and
haltOnFailure default to False, which doesn't make much sense.  I'll
change the default.  *However*, setting flunkOnFailure=True should
have helped, and in my experimentation it does help.  Can you
double-check that, and show us the configuration with that option that
doesn't flunk?

Dustin

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

Re: [Bug?] FileUpload more verbose on error

On Sat, Jan 1, 2011 at 12:16 PM,  <ariasgore@...> wrote:
> when I use FileUpload to store a file into a master's directory that does not exist it simply passes and does
not fail.
> So invalid paths do not create any stdio or stderr AND they simply pass.

I can't reproduce this - I added

  f1.addStep(FileUpload(slavesrc="README.txt",
masterdest="/tmp/nosuchdir/main.cp"))

and it created /tmp/nosuchdir for me successfully.  Changing to

  f1.addStep(FileUpload(slavesrc="README.txt",
masterdest="/usr/nosuchdir/main.cp"))

resulted in an exception (purple), which isn't ideal, but is certainly
not success (green).

> Btw and even if a FileUpload fails entirely it does not affect the Buildbot Status (and it has no flunkOnError)

I do see this, and will fix it up.

Thanks for the reports!

Dustin

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

Scheduler subclassing API

If you're subclassing schedulers in a substantive way, and you're
using Buildbot-0.8.x, please let me know - with pointers to your code
if possible.

I'm struggling with rewriting schedulers to use SQLAlchemy.  The
interface is not well-defined now, and depends pretty intimately on
the ad-hoc database interface used in 0.8.0 - 0.8.3.  It's also
horrendously inefficient - for every change, every scheduler wakes up
and makes several expensive database queries, when in most cases the
majority of the schedulers could determine trivially from the
in-memory Change object that they are not interested, and the
remaining schedulers could probably figure out what to do with one or
zero queries.

I'm going to have to break all but the simplest subclasses.  However,
I'm not ready to define and commit to a subclassing API for schedulers
(that's a 0.10.x feature -
http://trac.buildbot.net/milestone/0.10.%2B), and I want to
incorporate a message-passing architecture into the API.  So if you
have a nontrivial Scheduler subclass, you'll probably need to rewrite
it twice.  I'd like to minimize that pain.  Let me see what your code
is and I'll try to design things to make it simple.

Dustin

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

Re: Master Shell Command exist status does not affect overall build status

Hello,
retried again with flunk and it seems to work now.
Mayby cache or simply user-error.
That seems to work now.

Thanks
Sam

-----Ursprüngliche Nachricht-----
Von: djmitche <at> gmail.com [mailto:djmitche <at> gmail.com] Im Auftrag von Dustin J. Mitchell
Gesendet: Samstag, 1. Januar 2011 22:51
An: ariasgore <at> gmx.de
Cc: Amber Yust; buildbot-devel <at> lists.sourceforge.net
Betreff: Re: [Buildbot-devel] Master Shell Command exist status does not affect overall build status

On Sat, Jan 1, 2011 at 12:09 PM,  <ariasgore <at> gmx.de> wrote:
> I tried,
> no effect. Still everything green

I had a look, and MasterShellCommand's flunkOnFailure and
haltOnFailure default to False, which doesn't make much sense.  I'll
change the default.  *However*, setting flunkOnFailure=True should
have helped, and in my experimentation it does help.  Can you
double-check that, and show us the configuration with that option that
doesn't flunk?

Dustin

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl
_______________________________________________
Buildbot-devel mailing list
Buildbot-devel <at> lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/buildbot-devel

Re: [Buildbot] #1746: "<type 'exceptions.TypeError'>: sequence item 0: expected string, list found" errors

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl

_______________________________________________
Buildbot-devel mailing list
Buildbot-devel@...
https://lists.sourceforge.net/lists/listinfo/buildbot-devel