Re: [Libguestfs] [nbdkit PATCH 2/2] tests: Enhance captive test

On Mon, Sep 30, 2019 at 12:15:42PM -0500, Eric Blake wrote: > Test the just-fixed bug in --run failing to detect an nbdkit assertion > failure. > > While at it, sleep less when we don't actually need to wait for the > socket to be opened. > > Signed-off-by: Eric Blake <eblake(a)redhat.com&gt; > --- > tests/test-captive.sh | 46 +++++++++++++++++++++++++++++++++++++++---- > 1 file changed, 42 insertions(+), 4 deletions(-) > > diff --git a/tests/test-captive.sh b/tests/test-captive.sh > index e89c387d..88c0d818 100755 > --- a/tests/test-captive.sh > +++ b/tests/test-captive.sh > @@ -1,6 +1,6 @@ > #!/usr/bin/env bash > # nbdkit > -# Copyright (C) 2014-2018 Red Hat Inc. > +# Copyright (C) 2014-2019 Red Hat Inc. > # > # Redistribution and use in source and binary forms, with or without > # modification, are permitted provided that the following conditions are > @@ -36,13 +36,15 @@ set -x > > # Test nbdkit --run (captive nbdkit) option. > > +fail=0 > + > sock=`mktemp -u` > -files="$sock captive.out" > +files="$sock captive.out captive.pid" > rm -f $files > cleanup_fn rm -f $files > > nbdkit -U $sock example1 --run ' > - sleep 5; echo nbd=$nbd; echo port=$port; echo socket=$unixsocket > + sleep 1; echo nbd=$nbd; echo port=$port; echo socket=$unixsocket > ' > captive.out For some reason this test is now failing much more frequently under load. I can pretty easily reproduce the failure on my local 24 core machine, especially if I use ‘make check-valgrind’, and it happens in Koji too on the slower builders, eg: https://kojipkgs.fedoraproject.org//work/tasks/6571/38396571/build.log I am planning to revert this single line from the commit for now, but really I don't understand why it's failing because the sleep ought to be completely unnecessary (ie. no sleep ought to work just as well). I can't think right now why I put a sleep there in the first place ...