troubleshooting: Add instructions for enabling kdump #198
Conversation
|
It is probably simplest to ship it now and then hold the PR until it's reached stable, but you could add a note in the meantime to layer if it's not installed? |
| + | ||
| [source, bash] | ||
| ---- | ||
| sed -i "s/^path.*/path \/sysroot\/ostree\/deploy\/fedora-coreos\/var\/crash/" /etc/kdump.conf |
kelvinfan001
force-pushed
the
kfan-kdump-doc
branch
3 times, most recently
from
9005a29
to
7504fda
Compare
|
|
kelvinfan001
force-pushed
the
kfan-kdump-doc
branch
from
7504fda
to
3af11bb
Compare
|
#199 is in now |
kelvinfan001
force-pushed
the
kfan-kdump-doc
branch
from
3af11bb
to
ac86429
Compare
kelvinfan001
force-pushed
the
kfan-kdump-doc
branch
from
ac86429
to
d6e71d7
Compare
| + | ||
| [source, bash] | ||
| ---- | ||
| sudo rpm-ostree kargs --append='crashkernel=256M' |
There was a problem hiding this comment.
The doc suggests 128, maybe we should keep that 'default'? Unless there is a reason fcos needs more memory.
https://fedoraproject.org/wiki/How_to_use_kdump_to_debug_kernel_crashes
There was a problem hiding this comment.
Based off of a local qemu build of the current testing-devel, it seems like 128M does not work.
|
@dustymabe Looks like |
| @@ -0,0 +1,32 @@ | |||
| = Enabling kdump to debug kernel crashes | |||
There was a problem hiding this comment.
WDYT about rewording this as "Debugging kernel crashes using kdump" so that the title matches the menu item more closely?
| ---- | ||
| xref:kernel-args.adoc[More information] on how to modify kargs via `rpm-ostree`. | ||
|
|
||
| . Make sure the path in which the vmcore will be saved is somewhere in the `/var` directory of the FCOS stateroot. It is also possible to write the dump over the network; for additional information, consult the mkdumprd man page and the comments in /etc/kdump.conf. |
There was a problem hiding this comment.
| . Make sure the path in which the vmcore will be saved is somewhere in the `/var` directory of the FCOS stateroot. It is also possible to write the dump over the network; for additional information, consult the mkdumprd man page and the comments in /etc/kdump.conf. | |
| . Make sure the path in which the vmcore will be saved is somewhere in the `/var` directory of the FCOS stateroot. It is also possible to write the dump over the network; for additional information, see `mkdumprd(8)` and the comments in `/etc/kdump.conf`. |
| + | ||
| [source, bash] | ||
| ---- | ||
| sed -i "s/^path.*/path \/sysroot\/ostree\/deploy\/fedora-coreos\/var\/crash/" /etc/kdump.conf |
There was a problem hiding this comment.
Why can't we just use /var/crash?
There was a problem hiding this comment.
hmm you're right. I just tested and it seems like the default /var/crash seems to have no problems. I don't remember why I thought we had to modify it.
| sudo systemctl reboot | ||
| ---- | ||
|
|
||
| TIP: For additional information on how to test that kdump is properly armed and how to analyze the dump, refer to the https://fedoraproject.org/wiki/How_to_use_kdump_to_debug_kernel_crashes[kdump documentation for Fedora]. |
There was a problem hiding this comment.
Should we also link to https://www.kernel.org/doc/html/latest/admin-guide/kdump/kdump.html?
kelvinfan001
force-pushed
the
kfan-kdump-doc
branch
from
d6e71d7
to
57080ef
Compare
kelvinfan001
force-pushed
the
kfan-kdump-doc
branch
from
57080ef
to
d48385e
Compare
Since
kexec-toolsis currently not yet in the base image, I included a step to layerkexec-tools. Should I hold this PR until we've includedkexec-toolsin the base image?