[Crash-utility] what is recommended dump level (makedumpfile)

[Ken'ichi Ohmichi]

Hi,

Dave Anderson wrote:
>> what is recommended dump level when using makedumpfile
>> (kdump)? I'm looking for explanation like this.
>>
>> ------------ from diskdump document --------------------
>> Note that the partial dump feature has some risks.  There
>> are memory management lists which are scanned for a page's
>> memory attribute, so if the list has been corrupted, the
>> scanning process may fail.  For example, when specifying a
>> dump_level from 4-7 or from 12-15, the kernel's free page
>> linked lists are scanned; if the list is corrupt, diskdump
>> may hang.  Furthermore, it is possible that a page type
>> that has been skipped may be necessary to fully investigate
>> the cause of some issues.  Therefore, a memory collection
>> level should be selected to suit each situation.  The
>> recommended level is 19, because it is easiest to determine
>> whether a page is zero-filled or if it is a page cache page,
>> and because no page lists need to be traversed.
>> -------------------------------------------------------
>>
>> Does anyone know document describing these things
>> when using makedumpfile ?
>>
>>
>> Thanks,
>> toku.
> 
> I don't know of any document for makedumpfile other than the
> /usr/share/doc/kexec-tools-<version>/kexec-kdump-howto.txt
> and the output of "makedumpfile -h".  Neither of those has
> any caveats such as the diskdump document above, but I'm
> guessing that the same kind of issues would apply when makedumpfile
> walks page lists, the mem_map, etc.  But I don't know what the
> failure mode would be, given that it's a user application as opposed
> to a kernel module.
> 
> In any case, the makedumpfile maintainers are on this list
> and hopefully can shed some light on your query.

As Dave said, makedumpfile has the same risk as the partial dump
feature of diskdump. If excluding free pages and free pages linked
lists are corrupt, makedumpfile fails. So I recommend dump_level 15
(excluding zero pages, cache pages, user data pages. not excluding
free pages) because of the same reason as diskdump.


FYI:
If using RHEL5.2 or older, it is better that you update it to RHEL5.3.

RHEL5.2's makedumpfile has a buffer-handling bug and it cannot exclude
right page occasionally. This bug has been fixed on RHEL5.3, so if using
RHEL5.2 or older you should update it.

This bug has been fixed on makedumpfile version 1.3.0, and The following
thread is about this bug:
http://lists.infradead.org/pipermail/kexec/2008-September/002629.html


Thanks
Ken'ichi Ohmichi