Обсуждение: pgsql: Handle SIGTERM in pg_receivewal and pg_recvlogical
Handle SIGTERM in pg_receivewal and pg_recvlogical In pg_receivewal, compressed output is only flushed on clean exits. The reason to support SIGTERM as well as SIGINT (which is currently handled) is that pg_receivewal might well be running as a daemon, and systemd's default KillSignal is SIGTERM. Since pg_recvlogical is also supposed to run as a daemon, teach it about SIGTERM as well and update the documentation to match. While in there, change pg_receivewal's time_to_stop to be sig_atomic_t like it is in pg_recvlogical. Author: Christoph Berg <myon@debian.org> Reviewed-by: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com> Reviewed-by: Michael Paquier <michael@paquier.xyz> Discussion: https://postgr.es/m/Yvo/5No5S0c4EFMj@msg.df7cb.de Branch ------ master Details ------- https://git.postgresql.org/pg/commitdiff/8b60db774356117fab2eb53fb37160fa3e173cdb Modified Files -------------- doc/src/sgml/ref/pg_receivewal.sgml | 8 +++++--- doc/src/sgml/ref/pg_recvlogical.sgml | 18 ++++++++++++++++++ src/bin/pg_basebackup/pg_receivewal.c | 11 ++++++----- src/bin/pg_basebackup/pg_recvlogical.c | 9 +++++---- 4 files changed, 34 insertions(+), 12 deletions(-)
Daniel Gustafsson <dgustafsson@postgresql.org> writes:
> Since pg_recvlogical is also supposed to run as a daemon, teach it about
> SIGTERM as well and update the documentation to match. While in there,
> change pg_receivewal's time_to_stop to be sig_atomic_t like it is in
> pg_recvlogical.
While looking at this commit, I wondered why both of those programs
are declaring their signal handlers like
static void
sigint_handler(int signum)
rather than our standard convention
static void
pmdie(SIGNAL_ARGS)
Evidently we don't (any longer?) have any platforms where SIGNAL_ARGS
is non-default, but that still doesn't make this good coding. More
than once I've grepped for SIGNAL_ARGS to locate signal handlers.
Hmm, looks like same mistake in pg_waldump ... barring objection,
I'm going to run around and fix those.
I also notice that port.h has
typedef void (*pqsigfunc) (int signo);
extern pqsigfunc pqsignal(int signo, pqsigfunc func);
which is a bit inconsistent with the idea that SIGNAL_ARGS
might be different from that. Shouldn't we declare it as
typedef void (*pqsigfunc) (SIGNAL_ARGS);
?? I'm definitely going to do that for testing, because then
I can alter SIGNAL_ARGS to help me find any other stragglers.
regards, tom lane
> On 14 Sep 2022, at 17:06, Tom Lane <tgl@sss.pgh.pa.us> wrote: > > Daniel Gustafsson <dgustafsson@postgresql.org> writes: >> Since pg_recvlogical is also supposed to run as a daemon, teach it about >> SIGTERM as well and update the documentation to match. While in there, >> change pg_receivewal's time_to_stop to be sig_atomic_t like it is in >> pg_recvlogical. > > While looking at this commit, I wondered why both of those programs > are declaring their signal handlers like > > static void > sigint_handler(int signum) > > rather than our standard convention > > static void > pmdie(SIGNAL_ARGS) Sorry, I had missed that convention. > Evidently we don't (any longer?) have any platforms where SIGNAL_ARGS > is non-default, but that still doesn't make this good coding. More > than once I've grepped for SIGNAL_ARGS to locate signal handlers. > > Hmm, looks like same mistake in pg_waldump ... pg_test_fsync seems to have the same error. > barring objection, > I'm going to run around and fix those. No objections, unless you have better things to do and wan't me to clean up then I'll take care of it. > I also notice that port.h has > > typedef void (*pqsigfunc) (int signo); > extern pqsigfunc pqsignal(int signo, pqsigfunc func); > > which is a bit inconsistent with the idea that SIGNAL_ARGS > might be different from that. Shouldn't we declare it as > > typedef void (*pqsigfunc) (SIGNAL_ARGS); Agreed, that makes sense. -- Daniel Gustafsson https://vmware.com/
Daniel Gustafsson <dgustafsson@postgresql.org> writes:
> On 14 Sep 2022, at 17:06, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> barring objection,
>> I'm going to run around and fix those.
> No objections, unless you have better things to do and wan't me to clean up
> then I'll take care of it.
Nah, not your bug. There are a bunch of these, I'm finding ...
regards, tom lane
> On 14 Sep 2022, at 17:37, Tom Lane <tgl@sss.pgh.pa.us> wrote: > > Daniel Gustafsson <dgustafsson@postgresql.org> writes: >> On 14 Sep 2022, at 17:06, Tom Lane <tgl@sss.pgh.pa.us> wrote: >>> barring objection, >>> I'm going to run around and fix those. > >> No objections, unless you have better things to do and wan't me to clean up >> then I'll take care of it. > > Nah, not your bug. There are a bunch of these, I'm finding ... Thanks for tackling this! -- Daniel Gustafsson https://vmware.com/