DevHeads.net

boot issue with latest kernel

Hi all,

Has anyone seen this issue before? This afternoon, I tried updating a
bare metal CentOS 6 box, and got some odd error messages on the console
during booting kernel 2.6.32-754.11.1. (These aren't exact, I forgot
to try to get a photo of the console.)

sd 0:0:4:0 timed out resetting card
3w-sas timed out resetting card

Then the boot would simply hang, with no obvious disk activity on the
drives and no other messages on the console. Reverting back to an
earlier kernel (2.6.32-431.17.1) was perfectly fine. (Obviously this
is quite old hardware, but until today had never had problems.)

I noticed in the CentOS 6.10 changelog that 3w-sas has been deprecated,
but that it should still be supported. And even if 3w-sas had been
removed, I don't think that wouldn't explain timeouts on sd.

I have done a bunch of searches on the CentOS forums and on the web, but
didn't find anything specific to this issue with various combinations of
the error message keywords. If you have any pointers for me I would
greatly appreciate it!

--keith

Comments

Re: boot issue with latest kernel

By Young, Gregory at 03/20/2019 - 08:59

Hi Keith,

Is this a VM running on top of Hyper-V by chance? There seems to be an issue with the combination of:
- Windows 10 1708 or newer.
- Hyper-V config version 8.2 - 9.0.
- Microsoft's Fix for Spectre and Meltdown.
- Newer Intel Hardware (Sandy Lake/Kaby Lake, etc.)

The workaround seems to be to remove "rhgb" and "quiet" from the kernel line in the GRUB config (you may need to edit the command line from the grub menu before booting).

Greg

Hi all,

Has anyone seen this issue before? This afternoon, I tried updating a bare metal CentOS 6 box, and got some odd error messages on the console during booting kernel 2.6.32-754.11.1. (These aren't exact, I forgot to try to get a photo of the console.)

sd 0:0:4:0 timed out resetting card
3w-sas timed out resetting card

Then the boot would simply hang, with no obvious disk activity on the drives and no other messages on the console. Reverting back to an earlier kernel (2.6.32-431.17.1) was perfectly fine. (Obviously this is quite old hardware, but until today had never had problems.)

I noticed in the CentOS 6.10 changelog that 3w-sas has been deprecated, but that it should still be supported. And even if 3w-sas had been removed, I don't think that wouldn't explain timeouts on sd.

I have done a bunch of searches on the CentOS forums and on the web, but didn't find anything specific to this issue with various combinations of the error message keywords. If you have any pointers for me I would greatly appreciate it!

--keith

Re: boot issue with latest kernel

By Keith Keller at 03/22/2019 - 01:06

On 2019-03-20, Young, Gregory <gregory. ... at solarwinds dot com> wrote:
It is not; as I mentioned in my original post, it's bare metal.

--keith

Re: boot issue with latest kernel

By m.roth at 03/20/2019 - 11:42

Young, Gregory wrote:
mark

Re: boot issue with latest kernel

By Jonathan Billings at 03/20/2019 - 17:54

On Mar 20, 2019, at 11:42 AM, mark <m.roth@5-cent.us> wrote:

I agree with you, except for systems with a slow serial console. I leave ‘quiet’ there unless I’m debugging issues, because otherwise it *slows down the boot*, patiently waiting for the serial line to catch up as it prints a bunch of stuff I’ll see in dmesg anyway.

Re: boot issue with latest kernel

By Johnny Hughes at 03/20/2019 - 14:12

On 3/20/19 10:42 AM, mark wrote:
Good Plan :)