On 03/05/14 at 03:30pm, Jerry Hoemann wrote:
On Mon, Mar 03, 2014 at 06:37:16PM +0800, Baoquan He wrote:
> kdump now dumps vmcore to root partition by default in SAVE_PATCH
> directory, e.g /var/crash defaultly. This is problematic when another
> disk is mounted on /var or /var/crash, because the saved vmcore will
> he hidden after dump in 1st kernel. This also has the potential of
> blindly filling the root file system without a clue as to why.
>
> Now fix this by failing the loading of kdump kernel if dump target
> is root fs by default while different disk is mounted on save path.
>
> Signed-off-by: Baoquan He <bhe(a)redhat.com>
> ---
> mkdumprd | 24 ++++++++++++++++++++++++
> 1 file changed, 24 insertions(+)
>
--
Baoquan,
Thanks for making this change. I have hit this problem and this
modification will make things better.
This got me thinking about the issue, and I have an idea
for a next step and wanted your feedback.
Right now, fedora requires the pair { <partition>, <path> }
where path is relative to the root of the partition.
I find this a little confusing especially when using the gui
system-config-kdump.
As an alternative, could we instead of having the customer visible
entities that need to be configured as the above pair, could we
just have them specify the mounted path and walk up the path
to find the partition?
So, if they say "/home/user/crash" we first look if
"/home/user/crash"
is a mounted partition, if it is, this is the partition for crash
and our relative path is ".".
If crash isn't a mounted partition we check "/home/user" and if it is our
partition is that and the relative path is "./crash". And so on until
we get to "/".
I would say there's advantage for using only "path" option for filesystem
based dump, we can get all the informations including disk, mount option
etc. The only exception is ssh dump which can use old semantic.
But it is really too late for RHEL7.0 even RHEL7, so I think maybe we can
make it a future feature for RHEL8...
Thanks
Dave