Technical article 110337

I can't copy a system dump image bigger than 2GB on my OpenServer 5 system.

OpenServer 5.0.0OpenServer 5.0.2OpenServer 5.0.4OpenServer 5.0.5

Written · Last revised

Problem

My system panics and there is a system image in the dump device
(by default, this is /dev/swap).  When I reboot, I get prompted:

There may be a system dump memory image in the swap device. Do you want to save a copy of it? (y/n)

When I answer "y", a script directs me to save the dump image to my tape drive. However, when I transfer the tape contents to a file and run the crash(ADM) utility with the dump (crash -d dumpfile), I get errors and crash is unable to read the dump.

Various error messages from crash, accompanying this failure, may occur, such as:

Warning - <dumpfile> may not be a valid memory image

Read error on page directory entry at 0x2f00: Bad address (error 14)

Read error on bootinfo structure at 0x800

As an alternative, I attempt to run the sysdump(ADM) command, as in:

/etc/sysdump -i /dev/swap -n /unix -fumtominidump

See TA105935, "How do I create customized system dump images on demand?" for more details on using the sysdump command.

The result is essentially the same, however.  When I try to
run crash on the dump image (after re-inflation with sysdump)
I find that crash is unable to read the dump.

Running crash directly on the panic dump image in /dev/swap, however, as follows, is successful:

/etc/crash -d /dev/swap -n /unix

Cause

The script that is run in response ("y") to the above prompt,
/etc/dumpsave, invokes the ldsysdump utility.  This utility has
a problem correctly copying a system memory image from the swap
device that is larger than 2GB.
Another possible cause of the problem is the file size limit
in an OpenServer 5 filesystem (by default, the root filesystem
is an HTFS filesystem).  Because the maximum supported file size
in an HTFS filesystem is 2GB, copying a dump greater than 2GB
in size to a regular file is likely to fail.

The sysdump command has been reported to also have a problem with copying dump images from /dev/swap, if these images are greater than 2GB, even if the the final "mini" dump image is much less than 2GB in size.

Solution

These problems have been fixed in SCO OpenServer Release 5.0.6.

For earlier releases, the following workarounds can be used to successfully save the dump image generated by the panic to a tape device:

1. If you want to automatically save the dump to tape upon
panic, you can change the default dump device to a tape device.
In order to do this, see TA105920, "In OpenServer 5.0.x, I don't
have enough swap space to complete a panic dump."  Follow the
procedure in section II of this TA to configure your designated
tape device as the dump device.

2. If you want to copy the panic dump image from /dev/swap to a tape device, use the dd(C) command to copy the image:

  dd if=/dev/swap of=tapedev conv=bmode bs=1024 count=`expr \`memsize
/dev/swap\` / 1024 + 1`

where tapedev is the desired, configured tape device on the system, for instance /dev/rct0.

Having saved the dump image to tape, the dump can then be copied to a raw disk area for subsequent analysis with the crash utility.

From the Xinuos technical article database, record 110337. Reproduced with permission. The original is at www.sco.com/ta/110337, which is reachable but blocked from search engines. Xinuos marks this article against a retired product line. That refers to the product’s lifecycle, not to whether the advice still works — plenty of it still applies to running systems.

Next step

Need this fixed rather than explained?

Tell us the release, what the system runs and what it is doing. We will reply by email with what we would do next.

  • You deal with the people who do the work
  • No obligation, and no upgrade required before we will help
  • We tell you plainly if something is not worth doing

Send an enquiry

We reply by email. If a server is down, say so in the first line. You can also write to help@scoexpert.com.