Server Reboot

[kritlop]

We have an ongoing debate here at the office about how often our SCO 5.0.5 Server needs to be rebooted.&nbsp;&nbsp;Our office hours are from 8am-5pm Monday through Friday. We usually have 100 users on the system at once, and the total system memory is 512MB.&nbsp;&nbsp;I feel that if the server is operating properly, it shouldn't have to be rebooted.&nbsp;&nbsp;Does anyone have any thoughts on this?<br><br>Thanks

 

[edfair]

If you were a customer of mine, I would advise to take the system down every month or so, not because the system needs it, but to the reboot screen so the UPS can be unplugged. Or if you have automatiic shutdown, to test the automatic shutdown software by unplugging the UPS. <br>It also gives you a chance to make sure, in a test environment, that the emergency boot disk and data restore device are working correctly along with the normal cold boot procedures.<br>There's no thrill like having a system crash with 100 anxious users and finding that the emergency recovery stuff doesn't work.&nbsp;&nbsp;Actually 16 is the most I've had to listen to. And it was almost two months because they didn't listen to the advice.<br>And you are backed up, aren't you? And you do restore some files from the backup device to verify that it is readable, don't you?<br> <p>Ed Fair<br><a href=mailto: > </a><br><a href= > </a><br>

 

[zoonkai]

I agree with the previous response.....It would be good to take it down once a month to test emergency recovery etc.. However as you were asking...the system itself should be fine if you are not peaking out on performance.&nbsp;&nbsp;I know of an instance where a SCO 5.0.5 server was up for almost a year with heavy usage....and was working great....it was only shut down to move it to another room.&nbsp;&nbsp;Always backup and test the readability of your data....one of the best programs that I know of for this..( I use this one) is Lone tar (<A HREF="

http://www.cactus.com"

TARGET="_new">

http://www.cactus.com</A>)<br><br>Zoonkai

<p>Donald (Zoonkai) Dixon<br><a href=mailto:donnan@don-nan.com>donnan@don-nan.com</a><br><a href=

http://www.don-nan.com>Don-Nan

Pump & Supply Co.</a><br>

 

[pacman]

I tend to agree that &quot;if it ain't broke, don't fix it!&quot;&nbsp;&nbsp;Then again, it can't hurt to let cron do an autoreboot late at night to clean any frags or orphans.&nbsp;&nbsp;(Unix inherently likes lots of swap-file areas.)<br><br><br><br>Good Luck.<br>

 

[aquadan]

I would also concur.&nbsp;&nbsp;I am working on an account that has a SCO box that has not been taken down in over 2 years.&nbsp;&nbsp;It is working fine now.<br><br>A question for you would be, I just took this account, and I need to move the box.&nbsp;&nbsp;Any special procedures I should look at before I take the system down, and/or any scripts that need to be run prior to taking the system down?<br><br>Appreciate any help you may be able to give.

 

[edfair]

Have a backup. and verify that the backup is readable. Make sure you have install disks and documentation and any software/documentation that is required to get the backup device running. Make an emergency boot set and test. Get output of lpstat -t and a list of users. Document the hardware and make sure you know where to locate replacement parts. Systems are generally prone to failure on power-up more than while running. Then do it.<br>My record on a Xenix system, down once for a blown serial port in 8 years. Recabled a new location in 1992, moved it,&nbsp;&nbsp;took it down for the blown port, brought it back up and finally shut it down and scrapped it this year. OS isn't prone to breaking, unlike some others I know about. <p>Ed Fair<br><a href=mailto: efair@atlnet.com> efair@atlnet.com</a><br><a href= > </a><br>Any advice I give is my best judgement based on my interpretation of the facts you supply. <br>
Help increase my knowledge by providing some feedback, good or bad, on any advice I have given.<br>

 

[AndyBo]

I will strongly second the backup recommendation.&nbsp;&nbsp;In fact, take two backups, just in case the first one fails.&nbsp;&nbsp;A common problem on servers that have been running continuously for a long time is that the hard disk heads get stuck.&nbsp;&nbsp;What happens is that you power down the server, the hard disk heads park on a safe area away from your data.&nbsp;&nbsp;However, usually because heat has built up through continuous use, the heads semi-weld themselves to the parking area.<br><br>There are two cures that I know of.&nbsp;&nbsp;The first is to stay calm, and go and have a coffee.&nbsp;&nbsp;Give the disk at least fifteens minutes to cool off and then restart the server.&nbsp;&nbsp;Hopefully things will be OK.<br><br>If that doesn't work, the second option is to replace the disk &nbsp;&nbsp;That's why the backups are so important...<br><br>Hope everything goes OK for you. <p> <br><a href=mailto: > </a><br><a href= > </a><br>--<br>
0 1 - Just my two bits

 

[edfair]

AndyBo,&nbsp;&nbsp;not weld, stick.&nbsp;&nbsp;Has to do with lubrication on the platters. One way to free them up is to have out, rotate as fast as you can in the same plane as the platters, then stop the rotation with the heel of the hand. Platters will try to continue rotation when the frame stops. Have used successfully lots of times. No damage since the forces don't lift the heads. <p>Ed Fair<br><a href=mailto: efair@atlnet.com> efair@atlnet.com</a><br><a href= > </a><br>Any advice I give is my best judgement based on my interpretation of the facts you supply. <br>
Help increase my knowledge by providing some feedback, good or bad, on any advice I have given.<br>