Checklists for Maintenance Routine

[Sanjay_23]

I'm looking for maintenance tasks for my Unix box. Can you give me some tips on this on what should me included in the maintenance checklists?

[Sarvesh A Dhuri · Accepted solution]

Hi

for HP9000 servers, following commands are useful
1) dmesg ----> check for any error messages
2) check /var/adm/syslog/syslog.log file
3) uptime
4) check for logs by GSP login , use sl command for that. check attention led on generally you will found in HP 9000 A-class , N-class , L-class servers.
5) ps ----> check various process.
6) top ---> check for top process running on the systems , it shows priorities as well for every process , which will enable you to decide which process should carry higher priority than others.
7) bdf ---> check for the filesystem size.
8) ioscan ---> check whether system is able to communcate with peripheral devices.
variuos options are avialable with this commdn , check for man pages for ioscan.
9) use "ping" command for checking the network.
10) use "who -r" command , which will show in whis run level system is in.
if you have monitoring software such as glance plus , then its wonderful tool to monoitor system in terms of processors utilization , disk I/O , network information.

if servers are clusterd , then always check status using "cmviewcl" command.

[Printaporn_1]

perform make_recovery routinely.
check file system usage.

[lawrenzo]

you can use the Sar utility - I have attached a script that I downloaded from the itrc

also swapinfo is a good command to check swap space.

netstat with various options to see network stats

also ifconfig check the staus of the network card.

[Victor Fridyev]

Standard:
vgdisplay, swapinfo -ta,uptime, top -d 1 -f /tmp/top
Selfmade:
if [ $# -ne 0 ]; then
TRH=$1
else
TRH=96
fi
bdf -l | awk 'NR>1&&NF>1&&!/cdrom/ {if($(NF-1)>trh) print }' trh=$TRH

If you have DB servers, DB status is also interesting.

[SS_6]

Must run make_recovery before starting and do all necessary backus. Also prepare a backuot plan.You already got points for system logs etc.
clean the DDS drive.
Give a reboot if not done for a long time.
Can do Cluster failover test (If have Service Guard)
You can defrag file systems.
Look for latest Unix patch depot to install.
In the last I would recommend not to do all items at once which may impact system and troubleshooting . Make a list and priorities yor task as per downtime and need, many of the tasks can be done online.

[Sanjiv Sharma_1]

Hi Sanjay,

You should take the Ignite-UX backup, Full Filesystem Backup and your database/application backup running on the system before maintenence.

Run this nickel.shar script and ftp the output to your PC. It consists of all the useful information which may be required incase of your unix box failure.

# cd /tmp (If there is space problem you can do it in some other folder)
# sh nickel.shar.txt
# nickel
Mostlikely,
# /usr/contrib/bin/nickel

nickel.shar.txt script attached

hth.

[KCS_1]

Hi,


In the addition,the H/W log of the system by EMS if OnlineDignostic is installed.

# /var/opt/resmon/log/event.log


The LED status of the system that front and back panel of the box and pheripheral devices status LED.

[KCS_1]

/var/adm/shutdownlog (who -b)
/var/adm/swinstall.log

[Olivier Decorse]

i also check sometimes :
- /etc/rc.log and /etc/rc.log.old, to verify if any service don't start or stop correctly,
- run swverify \* for any patch/product not correctly installed (it takes sometimes),
- last -R to search for stranges/bad connexions
- any file in lost+found directories, probably lost data.
- any unavailable physical volume with pvdisplay,
- any stale Logicial Extent with lvdiplay -v.

Olivier.

[Shaikh Imran]

Hi,
I think you are asking maintenance task and not tasks which needs to be taken care for maintenance
First take a complete backup of your system
1) check the hard disks for any stale PE's
this equivalents to the bad blocks on the disk.
2) check the physical connectivity of all the network cards and connections
3) check the syslog file for any errors
4) check the mail for root and also
5) dmesg for some errors
6) check the tape drive for read-write-restore etc.,
7) check the CD-Rom drive for read and mount etc.,
8) If L,N or A class systems check the GSP logs or if older system check the chasis log
9) Install STM if not present and take the complete information of the hardware this will report if any problem in the hardware
10) STM verifier ( Licensing required) will test each hardware for the intermitent problems
11) check all the software installed are configured and running properly by swverify command ( see man severify for details)
12) check for the patches installed and latest patches required either individual or Bundle


Regards,