Hi!
Maybe need additional diagnostic?
02.02.2019, 20:41, "Evgeniy Efimkin" <efimkin@yandex-team.ru>:
> Hi!
> I reproduced it.
> 1. Make some write load on master: `pgbench -i -s 100000 testdb`
> 2. On sync replica drop output connections, while pgbench init database: `ip6tables -A OUTPUT -p tcp --dport 5432 -j
DROP`
> 3. Until was_sender_timeout reached, i send SIGHUB on master: `select pg_reload_conf();`
> After that walsender forever stay in WaitEventSetWait, because no timeout
>
> src/backend/libpq/be-secure.c:273
> WaitEventSetWait(FeBeWaitSet, -1 /* no timeout */ , &event, 1,
> WAIT_EVENT_CLIENT_READ);
--------
Efimkin Evgeny