Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Sat, Nov 17, 2018 at 2:27 PM Thomas Munro wrote: > Thanks for the review. Pushed. -- Thomas Munro http://www.enterprisedb.com

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Sat, Nov 17, 2018 at 5:56 AM Tom Lane wrote: > Thomas Munro writes: > > [ 0001-Add-WL_EXIT_ON_PM_DEATH-pseudo-event-v4.patch ] > > I took a quick look through this. I have no objection to the idea of > letting the latch infrastructure do the proc_exit(1), but I'm wondering > why this is in

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

Thomas Munro writes: > [ 0001-Add-WL_EXIT_ON_PM_DEATH-pseudo-event-v4.patch ] I took a quick look through this. I have no objection to the idea of letting the latch infrastructure do the proc_exit(1), but I'm wondering why this is in the thread that it's in. Is there any remaining connection

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

At Fri, 26 Oct 2018 14:13:51 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI wrote in <20181026.141351.09076928.horiguchi.kyot...@lab.ntt.co.jp> > Thank you for the fix. > > At Tue, 23 Oct 2018 17:26:37 +1300, Thomas Munro > wrote in > > > On Thu, Sep 6, 2018 at 9:57 PM Kyotaro HORIGUCHI >

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

Thank you for the fix. At Tue, 23 Oct 2018 17:26:37 +1300, Thomas Munro wrote in > On Thu, Sep 6, 2018 at 9:57 PM Kyotaro HORIGUCHI > wrote: > > And don't we need a description about this restriction in the > > function comment? > > Ok, added. Thank you. It looks good. > While

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Thu, Sep 6, 2018 at 9:57 PM Kyotaro HORIGUCHI wrote: > In sysloger.c, cur_flags is (just set but) no longer used. Right. Fixed. > === > In latch.c, > > - The parentheses around the symbols don't seem to be needed. > | (wakeEvents & (WL_EXIT_ON_PM_DEATH)) != 0 || > |

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

Ugrrr! PLEASE ignore this! It's not wrong at all. 2018年9月6日(木) 18:58 Kyotaro HORIGUCHI : > - The following assertion looks contradicting to the comment. > |/* Postmaster-managed callers must handle postmaster death somehow. */ > |Assert(!IsUnderPostmaster || > | (wakeEvents &

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

At Sun, 2 Sep 2018 07:04:19 +1200, Thomas Munro wrote in > > > # Is it intentional that the patch doesn't touch pgstat.c? > > > > Yes. pgstat.c still uses WL_POSTMASTER_DEATH because it does > > something special: it calls pgstat_write_statsfiles() before it exits. Mmm. Exactly.. > Rebased.

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Thu, Jul 19, 2018 at 11:51 PM Thomas Munro wrote: > On Thu, Jul 19, 2018 at 10:30 PM, Kyotaro HORIGUCHI > > Yeah. That seems good. Couldn't we reuse prepared WaitEventSet in > > other places? For example PgstatCollectorMain has the same > > characteristics, where WaitLatchOrSocket is used with

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Thu, Jul 19, 2018 at 10:30 PM, Kyotaro HORIGUCHI wrote: >> Hmm. Why wait any longer? The cluster is broken. Is there some >> correctness reason to defer shutdown in any of these places? > > Well, please don't get me wrong. I don't object to backends' exit > on postmaster death, rather I'm

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

At Thu, 19 Jul 2018 16:58:30 +1200, Thomas Munro wrote in > On Wed, Jul 18, 2018 at 8:30 PM, Kyotaro HORIGUCHI > wrote: > > At Wed, 18 Jul 2018 14:02:47 +1200, Thomas Munro > > wrote in > > > >> Here are some of the places I had to add WL_EXIT_ON_PM_DEATH: > >> gather_readnext(),

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Wed, Jul 18, 2018 at 8:30 PM, Kyotaro HORIGUCHI wrote: > At Wed, 18 Jul 2018 14:02:47 +1200, Thomas Munro > wrote in > >> Here are some of the places I had to add WL_EXIT_ON_PM_DEATH: >> gather_readnext(), shm_mq_send_bytes(), shm_mq_receive_bytes(), >> shm_mq_wait_internal(), ProcSleep(),

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Sat, Jul 14, 2018 at 5:34 AM, Heikki Linnakangas wrote: > On 18/04/18 09:55, Thomas Munro wrote: >> Here's a draft patch that does that. One contentious question is: >> should you have to opt *in* to auto-exit-on-postmaster death? Andres >> opined that you should. I actually think it's not

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On 18/04/18 09:55, Thomas Munro wrote: Here's a draft patch that does that. One contentious question is: should you have to opt *in* to auto-exit-on-postmaster death? Andres opined that you should. I actually think it's not so bad if you don't have to do that, and instead have to opt out. I

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Wed, Apr 18, 2018 at 6:55 PM, Thomas Munro wrote: > Here's a draft patch that does that. Here's a better one (the previous version could read past the end of the occurred_events array). -- Thomas Munro http://www.enterprisedb.com

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

Hi, I'd like to disentangle two related topics. For "I want PostmasterIsAlive() to go faster using signals on platforms that can support that", please see over here:

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Wed, Apr 11, 2018 at 12:47 PM, Thomas Munro wrote: > On Wed, Apr 11, 2018 at 12:26 PM, Andres Freund wrote: >> On 2018-04-11 12:17:14 +1200, Thomas Munro wrote: >>> I arrived at this idea via the realisation that the closest thing to >>>

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Wed, Apr 11, 2018 at 12:26 PM, Andres Freund wrote: > On 2018-04-11 12:17:14 +1200, Thomas Munro wrote: >> I arrived at this idea via the realisation that the closest thing to >> prctl(PR_SET_PDEATHSIG) on BSD-family systems today is >>

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

Hi, On 2018-04-11 12:17:14 +1200, Thomas Munro wrote: > I arrived at this idea via the realisation that the closest thing to > prctl(PR_SET_PDEATHSIG) on BSD-family systems today is > please-tell-my-kqueue-if-this-process-dies. It so happens that my > kqueue patch already uses that instead of

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Wed, Apr 11, 2018 at 12:03 PM, Andres Freund wrote: > On 2018-04-11 11:57:20 +1200, Thomas Munro wrote: >> Then if pgarch_ArchiverCopyLoop() and HandleStartupProcInterrupts() >> (ie loops without waiting) adopt a prctl(PR_SET_PDEATHSIG)-based >> approach where available as

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

Hi, On 2018-04-11 11:57:20 +1200, Thomas Munro wrote: > Rebased, but I don't actually like this patch any more. Over in > another thread[1] I proposed that we should just make exit(1) the > default behaviour built into latch.c for those cases that don't want > to do something special (eg

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Tue, Sep 20, 2016 at 11:26 AM, Andres Freund wrote: > On 2016-09-20 11:07:03 +1200, Thomas Munro wrote: >> Yeah, I wondered why that was different than the pattern established >> elsewhere when I was hacking on replication code. There are actually >> several places where

Re: [HACKERS] PATCH: Keep one postmaster monitoring pipe per process

On Thu, Sep 15, 2016 at 04:40:00PM -0400, Tom Lane wrote: > Thomas Munro writes: > > Very interesting. Perhaps that is why NetBSD shows a speedup with the > > kqueue patch[1] but FreeBSD doesn't. I guess that if I could get the > > kqueue patch to perform better