Hello list,
I'm not sure if my problem is due to updates, but this problem does coincide with the 2.6.14-1.1644_FC4 kernel update. On reboot, I noticed that my PostgreSQL database service fails. However, after I login to the Gnome desktop, I can execute /etc/rc.d/init.d/postgresql start in a conosle, and it works fine. What could cause this anomaly?
Thanks in advance for any suggestions.
Mike
On Wednesday 30 November 2005 13:23, Mike Leahy wrote:
Hello list,
I'm not sure if my problem is due to updates, but this problem does coincide with the 2.6.14-1.1644_FC4 kernel update. On reboot, I noticed that my PostgreSQL database service fails. However, after I login to the Gnome desktop, I can execute /etc/rc.d/init.d/postgresql start in a conosle, and it works fine. What could cause this anomaly?
I have a rather slow server box on which the postgresql init script always claimed that startup failed. However, postgresql was running quite happily with no apparent probs. I tracked it down to a timeout in the init script: /etc/init.d/postgresql, line 191-192: $SU -l postgres -c "$PGENGINE/postmaster -p '$PGPORT' -D '.blahblah......& ... sleep 2
The timeout was simply too short for postmaster to init itself on the old box (350MHz PII), and the wrapper script thought the whole thing had failed. I increased it to 6 seconds, and while the service takes a little longer to start (duh), it no longer displays an ugly red [FAILED] message.
As a side note, I had to do an SElinux relabeling to get postgresql to work initially.. Why, I don't know (the system has never run with selinux turned off). Relabeling seems to be the cure way too often, when strange problems occur that seem inexplicable (before you check /var/log/audit.log).
Øyvind
On 12/2/05, Øyvind Stegard oyvinst@ifi.uio.no wrote:
$SU -l postgres -c "$PGENGINE/postmaster -p '$PGPORT' -D '.blahblah......& ... sleep 2
The timeout was simply too short for postmaster to init itself on the old box (350MHz PII), and the wrapper script thought the whole thing had failed. I increased it to 6 seconds, and while the service takes a little longer to start (duh), it no longer displays an ugly red [FAILED] message.
On another note, Øyvind, I was experiencing the same issue and on the advice of your post upped the value of the timeout, however, I upped it to 3 seconds rather than 6. I'm running on a 550 MHz Celeron with 512MB of RAM. While I've not tested it after a reboot, I've not been able to reproduce the failure message with multiple repeitions of "service postgresql stop" followed by '"service postgresql start" nor with "service postgresql restart". You might be able to drop your timeout a few seconds and save some time.
I'll see what it does during a boot, but I suspect it will behave as expected.
-- Chris
"`The enemy we fight has no respect for human life or human rights. They don't deserve our sympathy,' he said. `But this isn't about who they are. This is about who we are. These are the values that distinguish us from our enemies.' - Sen. John McCain, R-Arizona
At 11:01 PM -0600 12/2/05, Christofer C. Bell wrote:
On 12/2/05, Øyvind Stegard oyvinst@ifi.uio.no wrote:
$SU -l postgres -c "$PGENGINE/postmaster -p '$PGPORT' -D '.blahblah......& ... sleep 2
The timeout was simply too short for postmaster to init itself on the old box (350MHz PII), and the wrapper script thought the whole thing had failed. I increased it to 6 seconds, and while the service takes a little longer to start (duh), it no longer displays an ugly red [FAILED] message.
On another note, Øyvind, I was experiencing the same issue and on the advice of your post upped the value of the timeout, however, I upped it to 3 seconds rather than 6. I'm running on a 550 MHz Celeron with 512MB of RAM. While I've not tested it after a reboot, I've not been able to reproduce the failure message with multiple repeitions of "service postgresql stop" followed by '"service postgresql start" nor with "service postgresql restart". You might be able to drop your timeout a few seconds and save some time.
Or, how about having a 2 second timeout test that loops up to 5 times, so that it will run fast if startup is fast? (An RFE might be appropriate if someone actually does this. Fixed timeouts, ugh.) ____________________________________________________________________ TonyN.:' mailto:tonynelson@georgeanelson.com ' http://www.georgeanelson.com/
On Saturday 03 December 2005 06:01, Christofer C. Bell wrote: <snip>
it to 3 seconds rather than 6. I'm running on a 550 MHz Celeron with 512MB of RAM. While I've not tested it after a reboot, I've not been
<snip> Yeah, well, I haven't bothered tweaking it, I don't restart the box very often =).
Øyvind