Re: Kmc might cause iva autopkgtest fail on arm64 (possibly memory issue?)
On Mon, Nov 16, 2020 at 09:53:05PM +0530, Nilesh Patra wrote:
> > > This is on ARM, and tests on amd64 are OK... could this just be a RAM
> > > issue on the test box? I very vaguely remember that kmc needs quite a
> > > bit of RAM, which I'm not sure all of the ARM CI boxes have?
> > > Just guessing.
> >
> > Its not only on arm, its also on i386[2]:
> >
>
> JTFR, I tried running the tests on a arm64 porter box, and I get the
> failure within a few seconds:
>
> $ /bin/bash test-example
> Running iva in test mode...
> Copied input test files into here: /tmp/tmp.3MwLHzYtUY/test
> Current working directory: /tmp/tmp.3MwLHzYtUY/test
> Running iva on the test data with the command:
> /usr/bin/iva --threads 1 --trimmomatic /usr/share/java/trimmomatic.jar
> --pcr_primers hiv_pcr_primers.fa -f reads_1.fq.gz -r reads_2.fq.gz iva.out
> The following command failed with exit code 1
> /usr/bin/iva --threads 1 --trimmomatic /usr/share/java/trimmomatic.jar
> --pcr_primers hiv_pcr_primers.fa -f reads_1.fq.gz -r reads_2.fq.gz iva.out
>
> The output was:
>
> The following command failed with exit code 139
> bash run_kmc.sh
>
> The output was:
>
> run_kmc.sh: line 2: 10022 Segmentation fault kmc -fa -m4 -k95 -sf1
> -ci25 -cs1000000 -cx1000000
> /tmp/tmp.3MwLHzYtUY/test/iva.out/tmp.common_kmers.zu2h8c0q/reads.fa kmc_out
> $PWD > /dev/null
>
> On checking the ram consumption:
>
> $ free -h
> total used free shared buff/cache
> available
> Mem: 11Gi 622Mi 281Mi 60Mi 10Gi
> 10Gi
> Swap: 15Gi 82Mi 15Gi
>
> Looking at this, it does not _feel_ like a RAM issue - maybe it is
> something else.
Thanks for checking. Once you have access to such a porter box, would
you mind running it via strace or gdb to get some more information?
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: