Newbie: How To find Core Dump files on Unix?! URGENT!

[Former Member]

Hi, i would like to know on How To find Core Dump files on Unix?!

I know they should be found in /usr/sap/<SYSTEM-ID>/<INSTANZ>/work

but there are no "core" files and also in tmp is nothing unusual but disk space is totally full.

So how to find the big files which i could delete to make system running again?!

can someone provide me with some infos?!

br

[Former Member]

If you are running out of space. You need to first determine which mount is filling up.

run a " df -k " to determine which mount point is at 100%. this listing will show you all the mount points on your UNIX box and

their utilization.

go to that mount point. for example if it shows that the mount point /usr/sap/SM1 is at or close to 100%. cd to that mount point

and run:

du -k |sort -nr +0|more

This will give you a listing of directories sorted by size. This list will start with the directories under your mount point and show the detail as you go deeper into the directory structure. That is the first directory listed will be the total for all directories under that one. As you go down the list, and into sub-directories the sizes will get smaller. From this list you can determine where space is being taken. Beware though that just deleting a file does not necessarily mean that you will get the space back. On Linux or Aix run an fuser -c, or fuser -u on the file to determine if a process has it. If a process has it you may not get the space back even if you issue an rm on the file.

Hope this helps.

[Former Member]

Hi, thx to u all!!!

But because of ur help now i am facing next question.

What kind of data can i delete so far?! Because i am no admin i have no clue which data the server needs. I know that i could delete core-files and logs, also some trc-files.

What else?! A collegue of mine told me that maybe because of patching the system might cause this problem that i could delete some jar's ?!

I have no idea?! Any suggestions again?!

THANK YOU!!

[former_member204746]

to find:

find / -name core

to delete :

find / -name core -exec rm {} \;

[Former Member]

Hi,

Sometimes the space crunch can occur due to core dumps. However, I would suggest you to check on the joblogs too.

The joblogs are usually created under the /sapmnt/<SID>/SYS/global/<client>JOBLG.

You cannot directly delete them on the OS level. But you can delete them by using the report RSBTCDEL.

Please refer note 6604 on how to perform the same.

Hope it helps.

Best Regards,

Srividya.R

[Former Member]

Hi, thx for ur responses.

Because i am really a noob i have some other questions:

1. which user i should use to search and destroy?! root-user or SYSID-User?!

2. I found no core files and the harddisk is still 100% full, what files might also cause this problem

3. What on database-side could cause this problems?! can i search for sth here.

4. i was unable to use the given scripts (noob!), what can i do else?!

Does someone have some HowTo on this issue?!

br - i really appreciate ur help!

[Former Member]

1. which user i should use to search and destroy?! root-user or SYSID-User?!

Always use the user with the least permissions to do the job, don't use root if your sidadm can do it. If you want to find / delete SAP files use sidadm.

2. I found no core files and the harddisk is still 100% full, what files might also cause this problem

In your first post you wrote that /usr/sap/SID/INST/work dir is full, it is most likely that some trace files got to large. Check for files like dev_*, dew_w0 for example is the trace file of work process 0, dev_disp is the trace of the dispatcher and so on. You either have to increase the size of the filesystem, or find the cause for the growing file. It can be due to an increased trace level.

3. What on database-side could cause this problems?! can i search for sth here.

This does not look like a database issue so far.

4. i was unable to use the given scripts (noob!), what can i do else?!

Which one, please post what you typed and the error you got.

Best regards, Michael

[markus_doehr2]

Did you try the "du" script?

Markus

[Former Member]

Do you have job logs? they do occupy a lot of space...please check on them.

Best Regards,

Srividya.R

[markus_doehr2]

You can try to write a small script like

cd /usr/sap

for i in `ls`; do du -hs $i; done

You will get a list of directories including the amount of KB/MB. Then you can step into that one and check again.

Markus