Saturday 24 June 2017

System dump for HP-UX v11.x operating system

In this post, I will cover one of the most entertaining and important topic  how we take system dump for HP-UX v11.x version operating system.

let's starting with V11.00, the following things have changed with respect to system dumps.

1. A dump does not necessarily contain all the memory pages.

2. Save core was replaced by 2 different commands:
  • Save crash for boot time dump image management.
  • Crashutil for port boot dump analysis.
3. Dump area can now be configured both in the kernel, or in the /etc/fstab file. Also, those dump areas don't need to be in the vg00 volume group.

Now we understand what is the role of Dump size.

Dump size:

Default dump size is quite small (not the full memory), but it should be ok for most usage. It can be modified if required using the crashconf utility.

Note: In crashconf , all sizes are in physical pages (4Kb on PA-RISC).

Now we will go for configuration part which is describe below.

Configuration:

Dump area/logical volumes must be created contiguous and with bad blocks relocation disabled. Use the following lvcreate options if creating manually (or use SAM ).

hpx:/># lvcreate -r n -C y ...

Configuring in the kernel:-

lvlnboot can be used to configure dump areas in the vg00 volume group, as long as you have a dump lvol line in the /stand/system file.

Run time configuration:-

Add lines in /etc/fstab with the following format:

device  /  dump  defaults 0 0

Using swap space as dump area:

Swap spaces can still be used as dump areas. In that case, both usage should be explicitly declared:

  • use lvlnboot twice ( - +-s +- and - +-v +- options) 
  • or put 2 lines in /etc/fstab (one for swap and one for dump type
When the system disk is mirrored, the swap LV is actually mirrored, whereas the same device used for dumping will only write to one disk.

Dump areas usage order:-

Dump areas are used in reverse order of their declaration, so it's important to declare last the dump areas not used for swap if you have some (this speeds up reboot, since swap can be activated before saving the dump image). 

Also, if your run time configuration does not include any dump area in the vg00 volume group, the kernel will first automatically use the primary swap space as dump device. This might slow down the reboot since the dump must be completely saved before reactivating this primary swap. To avoid that
  • declare non-swap dump partitions in /etc/fstab 
  • add a dump none line in /stand/system to explicitly use run-time configuration only 
Please comment on the post if you have any query regarding the system dump on HP-UX operating system.

No comments:

Post a Comment