From: Justin Pryzby <pryzby@telsasoft.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Hubert Zhang <zhubert@vmware.com>
Cc: tsunakawa.takay@fujitsu.com, pgsql-hackers@postgresql.org, Andreas Seltenreich <seltenreich@gmx.de>
Cc: Bruce Momjian <bruce@momjian.us>
Subject: Re: Multiple hosts in connection string failed to failover in non-hot standby mode
Date: Thu, 6 May 2021 11:26:51 -0500
Message-ID: <20210506162651.GJ27406@telsasoft.com> (raw)
In-Reply-To: <770711.1610318330@sss.pgh.pa.us>
References: <BN6PR05MB3492948E4FD76C156E747E8BC9160@BN6PR05MB3492.namprd05.prod.outlook.com>
<TYAPR01MB299012CAC510D2644AEC2246FE160@TYAPR01MB2990.jpnprd01.prod.outlook.com>
<BN6PR05MB3492DFB2FF32B771172CC2F2C9170@BN6PR05MB3492.namprd05.prod.outlook.com>
<229507.1610056206@sss.pgh.pa.us>
<770711.1610318330@sss.pgh.pa.us>
On Sun, Jan 10, 2021 at 05:38:50PM -0500, Tom Lane wrote:
> I wrote:
> > The problems that I see in this area are first that there's no
> > real standardization in libpq as to whether to append error messages
> > together or just flush preceding messages; and second that no effort
> > is made in multi-connection-attempt cases to separate the errors from
> > different attempts, much less identify which error goes with which
> > host or IP address. I think we really need to put some work into
> > that.
>
> I spent some time on this, and here is a patch set that tries to
> improve matters.
>
> 0001 changes the libpq coding rules to be that all error messages should
> be appended to conn->errorMessage, never overwritten (there are a couple
> of exceptions in fe-lobj.c) and we reset conn->errorMessage to empty only
> at the start of a connection request or new query. This is something
> that's been bugging me for a long time and I'm glad to get it cleaned up.
> Formerly it seemed that a dartboard had been used to decide whether to use
> "printfPQExpBuffer" or "appendPQExpBuffer"; now it's always the latter.
> We can also get rid of several hacks that were used to get around the
> mess and force appending behavior.
>
> 0002 then changes the reporting rules in fe-connect.c as I suggested,
> so that you might get errors like this:
>
> $ psql -h localhost,/tmp -p 12345
> psql: error: could not connect to host "localhost" (::1), port 12345: Connection refused
> Is the server running on that host and accepting TCP/IP connections?
> could not connect to host "localhost" (127.0.0.1), port 12345: Connection refused
> Is the server running on that host and accepting TCP/IP connections?
> could not connect to socket "/tmp/.s.PGSQL.12345": No such file or directory
> Is the server running locally and accepting connections on that socket?
>
> and we have a pretty uniform rule that errors coming back from a
> connection attempt will be prefixed with the server address.
52a10224 broke sqlsmith, of all things.
It was using errmsg as a test of success, instead of checking if the connection
result wasn't null:
conn = PQconnectdb(conninfo.c_str());
char *errmsg = PQerrorMessage(conn);
if (strlen(errmsg))
throw dut::broken(errmsg, "08001");
That's clearly the wrong thing to do, but maybe this should be described in the
release notes as a compatibility issue, in case other people had the same idea.
Clearing errorMessage during success is an option..
--
Justin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-hackers@postgresql.org
Cc: pryzby@telsasoft.com, tgl@sss.pgh.pa.us, zhubert@vmware.com, seltenreich@gmx.de, bruce@momjian.us
Subject: Re: Multiple hosts in connection string failed to failover in non-hot standby mode
In-Reply-To: <20210506162651.GJ27406@telsasoft.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox