Server takes too much time while rebooting

[praveen..]

Hi All,
My server is physically very far from me.
some times i needed to restart the server, so i used this command:

#shutdown -r

server went down but it took 4:30 hours to get up.

i think it is taking too much time and it effects on production.

what may be reason behind this and how can we resolve this issue?

please suggest,

and how can i check the boot logs?

Thanks

[Pete Randall · Accepted solution]

What model server is it? The old k boxes were notoriously slow but not four and a half hours slow. You can use the setboot command to adjust the so-called "SpeedyBoot" options to see if that helps. Do a man on setboot for details.


Pete

[Jaime Bolanos Rojas.]

Praveen,

Also you might want to check the syslog, to see if any service did not wanted to start and was just trying to start over and over again, that might be a reason why the server is taking such a long time.

Regards,

jaime.

[Patrick Wallek]

Check the /etc/rc.log file and see if there is anything there that indicates some sort of delay while booting.

This is the file that logs when everything is started by the /sbin/rc?.d/S* scripts. There are no real time stamps for each script, but there is a header and footer in the file to indicate when the run level transition was started and when it finished. This will give you someplace to start.

[DCE]

Factors that can affect boot time
Model of unit
Amount of memory
Amount of disk
Type of file system

Older systems (specifically the K class) take longer to boot.

extended meory checks take longer, the more memory installed.

The more cpu's the longer the cpu check

A system with all HFS file systems (usually an older system) will take a long time to boot because of the disk checking that is required.

If the system did not go down clean (or you have some questionable disks), there is the potential that a fsck is being run duringthe boot up. Depending on how many disks are affected this could take a long time. check your dmesg, syslog and rc.log for messages.

[Victor Fridyev]

Hi,

Did you try to connect to a console to see which part of reboot process takes most of the time ?
May be, you have a database on the machine and its shutdown requires disconnecting of all processes. So stop it before reboot.
Additional candidates are Service Guard, Databases, NFS servers etc.
If you don't have console access, add debug time printout to a file in /sbin/rc0.d and /sbin/rc3.d

HTH

[praveen..]

Model: ia64 hp server rx7620
RAM=32GB

couldnot find any error in /etc/rc.log

no oracle database is there, all the processes have stopped before rebooting.


I think i need to check 8 change the setboot options.

how to check the current values of setboot and how to change it?

please suggest

[Victor Fridyev]

Hi,

Run

#>setboot #in order to see current status
#>setboot -p primary_path
#>setboot -a alternate_path
#>setboot -s {on|off} #AutoSearch
#>setboot b {on|off} #AutoBoot

[freddy_21]

Is your boot disk using external storage? or internal storage.



thanks
freddy

[Torsten.]

If your boot options are wrong, most likely your system won't boot at all.

from Victors post:

"Did you try to connect to a console to see which part of reboot process takes most of the time ?"

Did you?

My first thought was an improper nfs setting. This could cause very long timeouts.

[Darren Etheridge_2]

I had a VERY similar problem with a rp7420 and found it was because of a problem with my NFS settings. Try connecting to the GSP and rebooting and watch for the NFS service to start, or try to start. After fixing this my reboots went from 45 minutes to 7. Good luck.

Darren

[inventsekar_1]

Hi Praveen,

* Let us know what softwares and applications running on that machine. that will help to find out where is the problem.

* by selecting the reqired services alone, you can speed up the boot.

[praveen..]

server is conected to external SAN but all other servers are also connected to that server.
and other server never took too much long time.

even i match the /etc/rc.log of two other servers with this server, both are same.

please suggest what to do?
should i use setboot option?

[freddy_21]

I think you must connect to console, and look at step by step booting the server. you will know what process take a long long time. Is it booting problem, NFS problem, or others.


thanks
Freddy

[Victor Fridyev]

Hi,

There may be an additional reaon for the case (thanks to SEP for the idea): a difference in configuration in /etc/hosts amd /etc/rc.config.d/netconf. In this case computer waits in NFS part. Additional reason is wrong NIS configuration. These both can produce some hours delay in startup process.
HTH

[Bill Hassell]

The setboot options will make a difference of a few minutes, not 4 hours. The long delay is definitely NIS and NFS which are both notorious is delaying bootup. If you aren't using NFS, turn everything off in the netconf file. The other technique is to add a date command for every script in /sbin/init.d. Now you can immediately see the bad script(s) in /etc/rc.log.