How to understand if a machine properly runs

[Enrico Venturi]

Hello colleagues,
I need a simple command, or a simple set of commands, to check if a machine is properly running, i.e. if it can support the start of an application.
I mean, something like "bdf" to check if the file system is full, some diagnostic on the HW to check if the disks are fault or not, and so on.
Is it possible to take a look at the system log? which are the strings that I should look for?

thanks
Enrico

[malay boy]

One tool I can think off is mstm or xtsm.A load of thing you could do from here.The one thing I always used is Disk execise.This is to check whether the disk is OK or Not.

still learning,
mB

[Ian Dennison_1]

It depends on what the application requires.

'swapinfo -ta' shows swap and memory allocation.
'bdf' does show disk space
'vmstat' interrogates various system tables.
'sar' also checks system tables.
'dmesg' shows system errors from a buffer.

If you are asking if you can write a set of scripts to perform capacity planning, the answer is probably not. You can't beat the analysis and experience of a good sysadmin for capacity planning.

Share and Enjoy! Ian

[john korterman]

Hi Enrico,
I would say it is impossible to guarantee that by executing a few commands. However, you can always do the following:
# vgdisplay -v | grep stale
If noting is returned, you probably have all your disks available, but it is not waterproof.
# top
for checking that you have some idle cpu time.
# vmstat
Check for the "po" (page out) column. If it says zero, the system does not swap.
Search /var/adm/syslog/syslog.log for the string "fail":
# grep -n fail /var/adm/syslog/syslog.log
which will give you the line number where "fail" appears; check that out - it is of course not a guarentee either, but many error messages contain the string "fail".
However, the best indicator is if the users do not complain!

regards,
John K.