Re: assertion at postmaster start - Mailing list pgsql-hackers

From Tom Lane
Subject Re: assertion at postmaster start
Date
Msg-id 10852.1560793134@sss.pgh.pa.us
Whole thread Raw
In response to Re: assertion at postmaster start  (Andres Freund <andres@anarazel.de>)
Responses Re: assertion at postmaster start
List pgsql-hackers
Andres Freund <andres@anarazel.de> writes:
> On 2019-06-15 12:09:50 -0400, Alvaro Herrera wrote:
>> Once in a blue moon I get this assertion failure on server start:
>> TRAP: FailedAssertion(«!(AbortStartTime == 0)», Archivo: «/pgsql/source/master/src/backend/postmaster/postmaster.c»,
Línea:2957) 

> And "server process" is afaict only used for actual backends, not other
> types of processes. But we've not yet seen "database system is ready to
> accept accept connections", so IIRC it could only be a "dead_end" type
> backend? But we didn't yet see an error from that...
> Seems to indicate a logic error in postmaster's state machine. Perhaps
> something related to dead_end processes?

So if Andres is guessing right, this must be from something trying to
connect before the postmaster is ready?  Seems like that could be
tested for ...

            regards, tom lane



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: initial random incompatibility
Next
From: Andres Freund
Date:
Subject: Re: UCT (Re: pgsql: Update time zone data files to tzdata release2019a.)