UCS Delay blades boot process after a chassis power cycle

[Presidio HS-Cloud]

Hi all,

we have some UCS blades that boot from SAN, in event of a power failure the blades complete the POST before the boot LUNs on the EMC VNX are available causing the blades to remain in "no boot device found ".

Is it possible to configure a maintenace policy or a bios policy to delay the boot process after a power cycle or fatal reset? Or to reset the server until it find a bootable device?

TIA

Roger

[Kirk J]

Greetings.

There are a couple of bios policy settings (boot options tab) that might allow you to accomplish this.

Cool Down Time, Number of retries, and Boot option Retry.

Thanks,

Kirk...

[Presidio HS-Cloud]

 Thank you Kirk,

I can't find this options, even if I change the Boot Option Retry to "enabled"...

If I expand "Bios Defaults" and select B200-M4, under the boot options tab I see

 Please Note
There are no applicable BIOS Settings in this category for the selected hardware model.If can help,

Cisco UCS B200 M4 blades (chassis 5108), FI6248UP.

[Presidio HS-Cloud]

Sorry duplicated reply sent...

can't find boot options UCSM 3.1.1 and 3.1.3

Regards

[Presidio HS-Cloud]

Hi, can anyone help?

TIA

Roger

[Kirk J]

My lab had 3.23d installed, so maybe that go introduced in 3.23 line.

Also, the defaults won't list this.

Try creating a bios policy, and look under the boot options tab, to see these related settings.

Kirk...

[Wes Austin]

This issue is being tracked already here:

https://bst.cloudapps.cisco.com/bugsearch/bug/CSCve28644/?reffering_site=dumpcr

Are you on fixed code?

[Presidio HS-Cloud]

Thank you Wes,

---

@Presidio HS-Cloud wrote:

Sorry duplicated reply sent...

can't find boot options UCSM 3.1.1 and 3.1.3

Regards

---

we are on 3.1(1h) and 3.1(3e)

[Wes Austin]

Sorry, overlooked that before replying. Yeah, I had pushed to have this fix implemented as you are not the only customer who has faced this same issue. :) We faced challenges initially in fear that if you have 20 hosts all try to login to storage over and over and over, it may cause bigger issues. However, we were able to implement this via BIOS token.

Please open a TAC case if you have any issues after upgrading to a version that supports this token.

[Presidio HS-Cloud]

Ok, so only versions 3.2.x have this option enabled? Instead of upgrading is it possible to make blades reset after a timeout configured in the Server Managment tab?

[Wes Austin]

If the OS watchdog timer works for your scenario, then you can use it. I would recommend testing to ensure the server will boot before the specified time in any failure scenario.

[Anadat]

Hi, I have tried to modify the policy and put watchdog timer works and it gives me the error in BIOS 8198.

Is there any solution to this problem?