Hi,
On 24.05.2013 10:21, Oliver-Rainer Wittmann wrote:
Hi,
On 23.05.2013 09:19, Oliver-Rainer Wittmann wrote:
[snip]
I have seen that #621 of aoo-win7 also had this error and that you had
again cleaned up the hanging processes.
In order to get some error output I will switch off the HTML output
(no
"--html" option) and the multiprocessor build (no "-P2 -- -P2"
options)
before the weekly clean build of aoo-win7 and for the aoo-w7ia2 build
which is always clean.
Build #113 of aoo-w7ia2 went well.
The failure occured just because I forgot to disable the collection of
the HTML build logs.
I'm confused - was there something wrong with the w7ia2 build. The
previous two builds were clean: http://ci.apache.org/builders/aoo-w7ia2
Other than stumbling over hung processes, I don't think there is an
issue with ia2.
We had observed the problem of the hanging process.
#105 had hanging process
#106 went well after your process cleaning
#107, #108, #109, #110 had again problems with hanging processes
#111 went well after your process cleaning
#112 the first successful build without your invention
#113 went well with my first temporary investigation stuff - failure
just because of error in copying non-existing HTML logs.
#114 went well with my second investigation stuff
I decided to do some investigation stuff as the problem showed up two
times in the last days. Currently, we do not hit the problem. But when
it occurs again, we will have some more information due to the different
logging. The problem of the current HTML logging is that we do not have
it, when the hanging process error occurs.
As #114 went well I will now switch on the default multiprocesses builds
- 2 modules built in parallel with each using 2 make/dmake processes.
#115 build of aoo-w7ia2 having standard multiprocessor builds (options
"-P2 -- -P2") went well.
I will leave the HTML output switched off until Monday. May be the
problem with the hanging process occurs. Then we have some output when
and where it occurs.
#117 build of aoo-w7ia2 showed again the error.
Thus, following #118 build fails - it could not remove
/ext_libraries/apr/wntmcsi12/misc/apr-1.4.5/Makefile.win
A deeper look into the #117 build output reveals the following:
- module apr was successful build, but the 'deliver' was not performed.
While searching for the reason Andre and myself again consulted Andrew's
information about the hanging processes. Andrew mentioned process
"cl.exe - cl /nologo /?". We search for the corresponding code which
starts such a process. We found it inside the makefile uses for the
Windows apr build - namely
/ext_libraries/apr/wntmcsi12/misc/apr-1.4.5/Makefile.win. The process
seems to be used to determine, if a 64bit compiler should be used or not.
As we have a 32bit compiler for our Windows platform, I will simplify
this makefile.
Hopefully this will solve our buildbot problems.
For the coming clean build of aoo-win7 at the weekend I will switch off
its HTML output in order to have some output in case the problem with
the hanging process occurs.
clean build of aoo-win7 went well. Thus, I reverted the temporary switch
off of the HTML output.
Best regards, Oliver.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
For additional commands, e-mail: dev-h...@openoffice.apache.org
