Automate system reboot

[david bernard_2]

Does anyone know how i could schedule OpenVMS to automatically restart?

I have tried creating a com file with the following 2 lines:
1.reboot :==@sys$system:shutdown 15 shutdown yes no later yes none
2.reboot

This seems to invoke the shutdown routine, but when it shuts down the queue's the job is ended before finishing.

I'm running OpenVMS v7.3-1 on a DS25.

Thanks in advance,
Dave

[Jim_McKinney · Accepted solution]

Submit the attached command procedure to batch to execute at the time of your choice. It will create a detached job that will execute the shutdown/reboot at that time.

[Jim_McKinney]

I notice that the file that I attached to my previous post had a line wrap. That double-quote character that you see alone on a line obviously belongs on the end of the previous command line.

[Bob Kirkwood]

Here is a procedure that I have been using for some years.

[Arch_Muthiah]

Dave,

Using SYSMAN's SHUT NODE command, you can do this.
$ mcr sysman
set env/node=target
shut node /automatic_reboot /invoke_shutdown -
/minutes_to_shutdown=mnts -
/reason="some" /reboot_check
exit
$ exit

Archunan

[Arch_Muthiah]

Dave,
The other method as per Mr.Jim suggestion, you can create one small comd procedure(st.com) with only command that is your shutdown command and RUN loginout.exe as DETACHED process...

st.com
-------
...
$ @sys$system:shutdown 10 "reboot" no - yes "auto reboot" no reboot_check
$ exit

and

$ run/detached/input=[]st.com -
/output=sys$manager:shutdown.log -
sys$system:loginout


Archunan

[Art Wiens]

Of course someone needs to ask ...

Why do you need to reboot VMS? ;-)

Cheers,
Art

[david bernard_2]

Thanks everyone!

Jim,

I tried the command you attached and it worked perfectly!!!

For some reason the system sometimes goes unresponsive - No connections allowed, which we're looking into, and testing our environment on a test system. In the meantime i've been restarting the system a few times a week in the middle of the night to suppress the issue.

Thanks again

[Ian Miller.]

when the system hangs then crash it and the crash dump can be analysed to determine the reason.

[comarow]

To repeat, with a forced crash, we can determine the cause.

Some of the most common causes are
page files filling up
$show mem/files
regularly

Running out of pagedyn
$show mem/full/pool
the pagedyn should be 50% free. It can't expand.

Not being able to expand npagedyn.
Seame
$sho mem/full/pool

Those account for probably 95% of the hangs.

Make sure you are up to date on ECOs.

[david bernard_2]

comarow,

I check the paged & nonpaged Dynamic Memory as you suggested.

Here's the output:
Paged:
Current Size = 7.94MB
Free Space = 4.67MB

NonPaged:
Current Size = 7.53MB
Maximum Size = 29.88MB
Free Space = 1.06MB

Should the nonpaged have more space available?

I know we are marginally behind on ECO's and have currently applied them to our test environment for observation, before installing in production.

Thanks,
Dave

[david bernard_2]

Another thing i have been attempting to do is monitor the % of CPU the processes on the system are using, but this seems to only be available for the top users.

Is there anyway to monitor the % of CPU use for all processes?

[Art Wiens]

A MONITOR SYSTEM/INT=1 will give you a total cpu percentage used indication.

MONITOR PROCESS/TOPCPU/INT=1 will show the top 8 (I believe it can fit 8 on the screen) if there are that many computing processes at the time.

Art

[labadie_1]

I guess this is a standalone node. If it is in a Cluster, from another node, you can do
sh sys/node=xxx, to see if you have processes in RWxxx states, or PFG, or MUTEX...
or
mc sysman set env/node=xxx
do ...
$ dir sys$examples:*mon*.com should show you 3 files which can have monitor stats run regularly. Then after a hang, you can see until when the node responded and if something strange happened, by replaying the monitor files.

$ monitor/input=file/beg=.../end=... monitor_item

But the best thing to do is to take crash, as already said. You will then know what is wrong and be able to cure that problem.