Re: cleanup log_disconnections()

From: Qingqing Zhou <zhouqq(at)cs(dot)toronto(dot)edu>
To: Neil Conway <neilc(at)samurai(dot)com>
Cc: pgsql-patches(at)postgresql(dot)org
Subject: Re: cleanup log_disconnections()
Date: 2005-10-03 16:28:26
Message-ID: Pine.LNX.4.58.0510031220240.13173@eon.cs
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-patches


On Mon, 3 Oct 2005, Neil Conway wrote:
> Why does the patch change behavior when
>
> - port->remote_port is the empty string

Ooops, this wording change is inconsistent with Log_connections, so I
reverted it. Updated patch is attached.

> - or, when end.tv_sec is negative
>

I can only imagine end.tv_sec becomes negative if the root changes system
time during the connection. We can't prevent this, and it is not our
problem to detect this (in fact, we can't), so just tell the truth
directly. If we detect this here, other places in the code should detect
this as well.

Thanks for pointing out, Qingqing

---

Index: postgres.c
===================================================================
RCS file: /projects/cvsroot/pgsql/src/backend/tcop/postgres.c,v
retrieving revision 1.463
diff -u -r1.463 postgres.c
--- postgres.c 26 Sep 2005 15:51:12 -0000 1.463
+++ postgres.c 3 Oct 2005 16:20:01 -0000
@@ -3526,29 +3526,13 @@
static void
log_disconnections(int code, Datum arg)
{
- Port *port = MyProcPort;
- struct timeval end;
- int hours,
- minutes,
- seconds;
+ Port *port = MyProcPort;
+ struct timeval end;
+ int hours,
+ minutes,
+ seconds;

- char session_time[20];
- char uname[6 + NAMEDATALEN];
- char dbname[10 + NAMEDATALEN];
- char remote_host[7 + NI_MAXHOST];
- char remote_port[7 + NI_MAXSERV];
-
- snprintf(uname, sizeof(uname), " user=%s", port->user_name);
- snprintf(dbname, sizeof(dbname), " database=%s", port->database_name);
- snprintf(remote_host, sizeof(remote_host), " host=%s",
- port->remote_host);
- if (port->remote_port[0])
- snprintf(remote_port, sizeof(remote_port), " port=%s", port->remote_port);
- else
- remote_port[0] = '\0';
-
gettimeofday(&end, NULL);
-
if (end.tv_usec < port->session_start.tv_usec)
{
end.tv_sec--;
@@ -3557,28 +3541,16 @@
end.tv_sec -= port->session_start.tv_sec;
end.tv_usec -= port->session_start.tv_usec;

+ /* for stricter accuracy here we could round - this is close enough */
hours = end.tv_sec / SECS_PER_HOUR;
end.tv_sec %= SECS_PER_HOUR;
minutes = end.tv_sec / SECS_PER_MINUTE;
seconds = end.tv_sec % SECS_PER_MINUTE;

- /* if time has gone backwards for some reason say so, or print time */
-
- if (end.tv_sec < 0)
- snprintf(session_time, sizeof(session_time), "negative!");
- else
-
- /*
- * for stricter accuracy here we could round - this is close
- * enough
- */
- snprintf(session_time, sizeof(session_time),
- "%d:%02d:%02d.%02d",
- hours, minutes, seconds, (int) (end.tv_usec / 10000));
-
- ereport(
- LOG,
- (errmsg("disconnection: session time: %s%s%s%s%s",
- session_time, uname, dbname, remote_host, remote_port)));
-
+ ereport(LOG,
+ (errmsg("disconnection: session time=%d:%02d:%02d.%02d,"
+ " user=%s, database=%s, host=%s, %s%s",
+ hours, minutes, seconds, (int) (end.tv_usec / 10000),
+ port->user_name, port->database_name, port->remote_host,
+ port->remote_port[0] ? "port=" : "", port->remote_port)));
}

In response to

Responses

Browse pgsql-patches by date

  From Date Subject
Next Message Kouber Saparev 2005-10-03 18:28:37 Re: BUG #1927: incorrect timestamp returned
Previous Message Dave Page 2005-10-03 15:07:28 Re: [HACKERS] External Sort timing debug statements