On 2021-10-27 21:15+0000, Eric Wong wrote:
> Thomas Weißschuh <thomas@t-8ch.de> wrote:
> > On 2021-10-26 05:28+0000, Eric Wong wrote:
> > > Thomas Weißschuh <thomas@t-8ch.de> wrote:
> > > > The only failures I still see are in t/lei-q-remote-import.t which is unrelated
> > > > to the git branch and some more that fail only during parallel test execution.
> > > > 
> > > > In t/lei-q-remote-import.t it seems the search results are not memoized to the
> > > > local store. (See attachment)
> > > 
> > > Thanks, I'll check out lei-q-remote-import.t separately.
> > > That could be related to SMP or slow/fast storage.
> > > Any details you can share about CPU core count, speeds
> > > or storage speeds, or CPU scheduler, CONFIG_HZ?
> > 
> > CPU:
> > 
> >   Model name:            11th Gen Intel(R) Core(TM) i7-1165G7 @ 2.80GHz
> >     CPU family:          6
> >     Model:               140
> >     Thread(s) per core:  2
> >     Core(s) per socket:  4
> >     Socket(s):           1
> >     Stepping:            1
> >     CPU max MHz:         4700,0000
> >     CPU min MHz:         400,0000
> > 
> > Storage: Samsung EVO 970 (with plenty of memory cache)
> > CONFIG_HZ: 300
> 
> I'm puzzled by this failure; I'm not sure if I've
> ever seen it.  Does it fail when run standalone?
> (prove -bvw t/lei-q-remote-import.t)

It also does. The protocol is attached.

> Maybe running under TEST_LEI_ERR_LOUD=1 or checking
> syslog can reveal something.
> 
> Adding a sleep shouldn't be necessary, but maybe something
> else is broken I'm not seeing...:

It didn't help, neither did larger sleeps.

> diff --git a/t/lei-q-remote-import.t b/t/lei-q-remote-import.t
> index 92d8c9b6058c..a2d643cd06c7 100644
> --- a/t/lei-q-remote-import.t
> +++ b/t/lei-q-remote-import.t
> @@ -35,6 +35,7 @@ test_lei({ tmpdir => $tmpdir }, sub {
>  	lei_ok(@cmd, '-I', $url);
>  	is_deeply($slurp_emls->($o), $exp1, 'got results after remote search');
>  	unlink $o or BAIL_OUT $!;
> +	sleep 1;
>  	lei_ok(@cmd);
>  	ok(-f $o && -s _, 'output exists after import but is not empty') or
>  		diag $lei_err;

I'll see if I can debug it in the coming days.