OS Initiated Reboot - Intermittently Fails - Gets to GRUB then KVM goes "black" and OS does not load

[troyheiss]

Hi All,

I'm getting up-to-speed with our Cisco UCS environment as we've been an HP and Dell shop for years.

I've installed SLES 11SP3 on a B200 M4 and I'm having some intermittent issues with OS initiated reboots.  The result is that the KVM console may go completely “black” right after the GRUB menu displays the kernel/os to load and then *poof*, the box is non-responsive meaning the OS doesn't load.  The only way to get the system back online is to powercycle the Blade once or twice.

Is it possible to actually move the HDD's to another physical blade and effectively move the assigned profile to that box and ensuring that WWNs move to that blade?  The idea is to see if it's maybe some crazy local hardware issue.

Thank you for your time.

[Brad_80]

Yes, you absolutely can migrate the local drives and service profile to another blade to rule out a hardware issue. As a precaution, you will want to ensure the scrub policy associated with the profile is set to not scrub the disks on disassociate. This policy can be changed live without requiring a reboot of the blade. Once you have verified the scrub policy will not be destructive, you can disassociate the profile, move the disks to the new target blade, and associate the profile.

[troyheiss]

Thank you for the specifics involved and the caution about the Scrub policy!

After verifying "Disk Scrub" was set to "no," I then Disassociated the profile from Server A, moved the two disks(RAID 1), to Server B, Associated the profile and powered up the Server successfully.

Turns out that the issue was an OS bug:

https://www.suse.com/support/kb/doc?id=7017498

booting is delayed and the system may appear to hang for between 30 to 60 minutes, if left long enough the system may eventually boot normally.