topic Re: Warning message in Operating System - HP-UX

Warning message

augusto cossa — Thu, 17 Aug 2000 08:43:56 GMT

Hi,

I have rebooted the server and after running dmesg command here found I the above warning message:

WARNING: Insufficient space on dump device to save full crashdump.
Only 536870912 of 536871936 bytes will be saved.
Dump device table: (start & size given in 1-Kbyte blocks)

What should I do?

Thanks,
Augusto

Re: Warning message

Patrick Wessel — Thu, 17 Aug 2000 08:50:41 GMT

Figure out in which filesystem the dump is copied and free up some space. Usually this is in /var (/var/adm/crash) But you are able to find the path in /etc/rc.config.d/savecore (with hp-ux 11.0 it's savecrash)

Re: Warning message

John Palmer — Thu, 17 Aug 2000 08:53:01 GMT

Don't worry about it. The system is warning you that your primary swap/dump volume is slightly smaller than the amount of main memory that you have.

This means that not all memory could be dumped in the event of a system panic.

Re: Warning message

Patrick Wessel — Thu, 17 Aug 2000 08:53:11 GMT

my fault, Stefan is right!
sorry

Re: Warning message

Stefan Farrelly — Thu, 17 Aug 2000 08:54:26 GMT

you dont have enough swapspace to dump all of your system memory in the event of a system crash. If you have say 1Gb of RAM then you need to have at least 1Gb of swapspace for it to dump to. If you already do then ensure you have this entry in your /stand/system file; dump lvol (if not add it in and rebuild your kernel and reboot)

Re: Warning message

Victor BERRIDGE — Thu, 17 Aug 2000 09:03:10 GMT

I agree with John, since the 3 main filesystems of vg00 have to be on 2 GB contiguous, and that you can have lots of RAM...

Re: Warning message

John Palmer — Thu, 17 Aug 2000 09:09:53 GMT

You could also upgrade to HPUX 11 in which version the dump strategy has been changed so that not all of memory is saved.

Only when your server crashes and you need to send the crashdump to HP for analysis do you need the full amount of dump space. I suspect that this is not likely to be often, if at all.

Re: Warning message

augusto cossa — Thu, 17 Aug 2000 09:11:31 GMT

Stefan,

I have vi /stand/system file and here is:
* Kernel Device info

dump lvol

How can I increase the swap device?

Thanks

Re: Warning message

augusto cossa — Thu, 17 Aug 2000 09:12:05 GMT

Stefan,

I have vi /stand/system file and here is:
* Kernel Device info

dump lvol

How can I increase the swap device?

Thanks

Re: Warning message

Trevor Dyson — Thu, 17 Aug 2000 09:21:38 GMT

Augusto,

If you wish to increase swap space to allow a full dump of main memory then the next steps depend on whether you are using HP-UX 10.x or 11.x

If you are using HP-UX 11.x then you should have a look at the crash dump white paper at the following URL:

http://docs.hp.com/hpux/content/syscrash.html

If you are using HP-UX 10.x then please let us know.

Regards, Trevor

Re: Warning message

CHRIS_ANORUO — Thu, 17 Aug 2000 09:25:21 GMT

Hi Augusto,
You can create another swap device on vg00 maybe 1Gb. Go throw SAM, Disk and Filesystem/Swap and action to Add Device Swap

Re: Warning message

augusto cossa — Thu, 17 Aug 2000 09:51:14 GMT

Trevor,

I'm using HP-UX 10.xx

Thanks

Re: Warning message

Trevor Dyson — Thu, 17 Aug 2000 12:05:38 GMT

Hi Augusto,

I am a bit rusty on HP-UX 10.x dump configuration but here goes (can someone else check out my logic?)

Firstly I think Chris's suggestion (previous post to this one) of increasing swap space will not increase dump space as only primary swap is used by default for dumps. The primary swap is the first swap area enabled by the kernel on boot up. If Chris is suggesting that you replace primary swap with a larger logical volume then this will work but it might take a bit of effort to set up.

Here is how I would tackle your problem. The warning message says "Only 536870912 of 536871936 bytes will be saved." This is only 1024 bytes short of what is required. The experts say that you should make sure that you have sufficient dump space to hold all of main memory plus a bit more for overhead for crash dump structures. The impact of not having enough dump space is that you will end up with an incomplete crash dump image that may not contain the necessary information to assist in analysing the reason for the system crash. As you are only 1024 bytes short I believe it is highly unlikely that if a crash dump were created and analysed that the problem could not be resolved just as easily as if you had a full crash dump. I would probably just ignore this.

However if I had some spare time on my hands I might fix this up (I hate loose ends) so here is what I would do.

Create a new logical volume of the smallest possible size in the root volume group. The new logical volume must be within the first 2GB of the disc on which it is created, it must be contiguous and have bad block relocastion turned off. Here is an example command:

lvcreate -l 1 -C y -r n -n lvdump vg00

Next go into SAM and got to Kernel Configuration -> Dump Devices to enable the lvdump logical volume as additional dump space. You should be able to get SAM to find the new logical volume to use as dump, I can't remember the exact menu options but from memory it is pretty straight forward.

If a crash were to then occur, the lvdump logical volume and primary swap space would both be used to contain the crash dump.

Regards, Trevor

Re: Warning message

Trevor Dyson — Thu, 17 Aug 2000 12:18:55 GMT

....one other thing.

If you have additional swap areas (called secondary swap) enabled the you can use these for dump space as well. Let's assume that your primary swap is on lvol2 and your secondary swap is on lvol9 in the root volume group. You can enable lvol9 as a dump device by doing the following:

lvlnboot -d lvol9 /dev/vg00

then reboot.