On Tue, 24 Sept 2024 at 04:57, Paul Gevers <elb...@debian.org> wrote:
> It took quite some manual iterations, but it failed in the end. > > root@elbrus:/tmp/autopkgtest-lxc.lyg9_rfi/downtmp/build.aKs/src# grep -E > '^(Buffers|Cached|SReclaimable):' /proc/meminfo ; vmstat > Buffers: 0 kB > Cached: 280176 kB > SReclaimable: 0 kB > procs -----------memory---------- ---swap-- -----io---- -system-- > -------cpu------- > r b swpd free buff cache si so bi bo in cs us sy > id wa st gu > 20 0 0 263208988 0 280176 2074908 9521847 15275564 14671816 > 180601581 7 20 4 75 0 0 1 > OK, I've realised this is an autopkgtest failure, not an upstream test failure. The error is: 48s test_no_arguments 48s [1;31mASSERT: [0mvalue line 48s [1;31mshunit2:ERROR [0m test_no_arguments() returned non-zero return code. Importantly, its test_no_arguments() that is non-zero, not vmstat. Which is from this: grep -E '^procs[ -]+memory[ -]+swap[ -]+io[ -]+system[ -]+cpu[ -]+$' "${stdoutF}" > /dev/null assertTrue 'header line' $? grep -E '^ +r +b +swpd +free +buff +cache +si +so +bi +bo +in +cs us sy id wa st gu$' "${stdoutF}" >/dev/null assertTrue 'header second line' $? grep -E '^([[:space:]]+[[:digit:]]+){18}$' "${stdoutF}" >/dev/null assertTrue 'value line' $? The "value line" is failing because the line is not 18 numbers, but 10, then 8. If you append the 1806... number to the previous line, you get 81 characters for that line. So, I think we have something in shunit2 not liking long lines and failing (sometimes) because of this. The fix is probably to adjust the autopkgtest grep to look for only a few numbers, not 18. That's a bit annoying but when the identical test in DEJAGNU *always* passes, its the best that can be done. - Craig > > Paul > >