Re: t/cindex.t "associate w/o search" test hangs for me

[Eric Wong]

Konstantin Ryabitsev <konstan...@linuxfoundation.org> wrote:
> On Tue, Nov 14, 2023 at 10:16:53PM +0000, Eric Wong wrote:
> > Konstantin Ryabitsev <konstan...@linuxfoundation.org> wrote:
> > >          └─-cindex -u --al,4432
> > >              ├─cidx shard[0],4646
> > >              └─cidx shard[1],4647
> > > 
> > > Anything I can do to figure out why this is happening?
> > 
> > You can show me strace and lsof +E of the processes (any other
> > processes (join|sort|awk|perl)?).  This code is highly in flux,
> > so it's also fine to rm the test for now since nothing
> > public-facing is using -cindex, yet...
> 
> Yeah, but I figured I'll poke a bit in case it's helpful.
> 
> I can't do +E because that's not available to me under CentOS7 (I can't wait
> until we move on, but just when we think the yak is fully shaved, we find more
> clumps of thick fur we hadn't considered). Is the output of the regular "lsof
> -p" helpful at all?

Sure.

> Strace for all three processes (-cindex, cidx shard[0], cidx shard[1]) just
> sits at:
> 
>     select(24, [13 16], NULL, NULL, NULL

OK, that's still useful.  One FD is signalfd, the others would
be a SOCK_SEQPACKET socket, I think...

> As far as I can see, there are no other processes other than cidx.

OK.  Hmm.. Perhaps `kill -CHLD' on the top-level cindex process
can move it along?  There's still some weird timeout/sleep
behavior leftover from Danga::Socket that should probably be
removed:

-------8<------
Subject: [PATCH] ds: run @post_loop_do if any user-queued events run

---
 We can probably kill more hacky wakeup behavior throughout
 cindex and maybe other places with this patch...

 lib/PublicInbox/DS.pm | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/lib/PublicInbox/DS.pm b/lib/PublicInbox/DS.pm
index da26efc4..4c8b502f 100644
--- a/lib/PublicInbox/DS.pm
+++ b/lib/PublicInbox/DS.pm
@@ -144,13 +144,14 @@ sub next_tick () {
                # https://rt.perl.org/Public/Bug/Display.html?id=114340
                blessed($obj) ? $obj->event_step : $obj->();
        }
+       1;
 }
 
 # runs timers and returns milliseconds for next one, or next event loop
 sub RunTimers {
-       next_tick();
+       my $ran = next_tick();
 
-       return ($nextq ? 0 : $loop_timeout) unless @Timers;
+       return ($nextq || $ran ? 0 : $loop_timeout) unless @Timers;
 
        my $now = now();
 
@@ -159,10 +160,11 @@ sub RunTimers {
                my $to_run = shift(@Timers);
                delete $UniqTimer{$to_run->[1] // ''};
                $to_run->[2]->(@$to_run[3..$#$to_run]);
+               $ran = 1;
        }
 
        # timers may enqueue into nextq:
-       return 0 if $nextq;
+       return 0 if $nextq || $ran;
 
        return $loop_timeout unless @Timers;