There is no disk space to dump the complete core file

[dong_1]

In case. There is no disk space to dump the complete core file. Can you show me method to resolve it.
Thanks for your helping me
Take care.

[Michael Tully]

You can either increase the size of /var (/var/adm/crash) which is the default, or create another logical volume in which to use. Make sure that you have around 50% of your RAM available as disk mounted as a filesystem. System core dumps do get compressed.

Have a look in /sbin/init.d/savecrash /sbin/init.d/crashconf and /etc/rc.config.d/crashconf and /etc/rc.config.d/savecrash files

[Sundar_7]

Hi,

You have a seperate dump device or using the swap device as the dump device ??

It is always recommended to have a seperate dump device file. If swap device is also used as the dump device and if the system is rebooted once without saving the core dump then it happens that the startup script enables the swap on the swap device using swapon and you cannot save the core dump on the filesystem any more.

Sundar.

[Sridhar Bhaskarla]

Hi,

There are two things to look at for system crash core files.

1. Enough disk space to save the dump initially during the system crash. You can find it out with 'crashconf -v' command. Look at "Total pages included in dump" (say dump_needed) and multiply with 4K to get the required space in KB. Compare it with the total that is displayed under the DEVICE section (say total_dump_configured). If total_dump_configured is less than dump_needed, then you will need to add more dump space to the system.

2. Once the system dumps the image successfully, then it will be saved to a filesystem while the system boots up. You will need to have atleast 1/3rd of the space in /var/adm/crash (if you didn't customize it) as the dump will be compressed before it is saved. If you don't have enough space in crash directory, then the savecrash process will fail. You may not be able to retrieve the dump later if this dump area is from swap.


-Sri

[Bill Hassell]

A system crash dump takes place in two steps. The first is to write out the contents of memory to the dump area. Typically, this dump area is shared with swap areas so as long as you have enough swap space on vg00 to hold all of RAM, you'll be OK for step 1.

Next the system reboots and after the filesystems are mounted, startup will look at the configuration in /etc/rc.config.d/savecrash to see if a crash dump is to be taken (SAVECRASH=1) and what options are to be used. The crash will normally be copied from the dump area to /var/adm/crash (per the savecrash file) but can be changed. Or, you can turn off saving the crashdump with SAVECRASH=0. If you don't have disk space, this is the easiest thing to do.

Of course, the best thing to do is to prevent crashes by regular patching, but if one does occur and you cannot save the results to disk, you can run savecrash manually with the -t option to save the crash to tape. You can then have the tape analyzed at a later time.