Topic: SME 7.4 clock and APIC and USB

[RBT]

I have always had trouble with the clock in SME and have found that using the noapic directive in grub.conf will fix the clock so it keeps time.

However, with SME 7.4, I still need my noapic directive for the clock, but that directive now stuffs USB and USB drives won't accept an address.  Remove the noapic directive and USB works, the drives accept an address and can be mounted, but the clock skips ahead badly. Catch 22.

Why do I need the noapic directive to make my clocks run on time? All my machines are AMD boxes.

This is not related to virtual machines at all, they are all hardware boxes.
 

[David Harper]

Welcome to the forums

However, with SME 7.4, I still need my noapic directive for the clock, but that directive now stuffs USB and USB drives won't accept an address.

This sounds like a bug. Please open a new bug in the tracker, and then post the number here as a reference.

[RBT]

Can do, but do you know why the noapic directive fixes the clock in the first place.

Oh, and can you link me to your bug tracker please.

[Stefano]

hi

please raise a bug for it.. as SME uses the CentOS/Redhat kernel, I guess it's somethig that could be corrected from CentOS/Redhat devs.

ciao
Stefano

[David Harper]

According to LinuxQuestions, some motherboards aren't built to the exact specifications required to meet the APIC standard. Thus some features might not work all that well.

As APIC was developed by Intel, it makes sense that AMD processors would be more likely to be affected. (Slightly OT, but... I learned my lesson with the K6-2: Never Again.)

EDIT: Click "Bugs" on the top menu to access the tracker.

[RBT]

So if I use Intel motherboards I won't need to use this option?

[David Harper]

Not necessarily; some Intel MBs need "noapic" as well. But "noapic" causing USB to fail... that's got to be related to the way the kernel and the motherboard are interacting. If it were happening on all motherboards Red Hat would have heard the screaming months ago and fixed the issue upstream.

[RBT]

Mar 30 09:04:15 server kernel: usb 6-3: new high speed USB device using address 2
Mar 30 09:04:26 server kernel: usb 6-3: device not accepting address 2, error -110
Mar 30 09:04:26 server kernel: usb 6-3: new high speed USB device using address 3
Mar 30 09:04:36 server kernel: usb 6-3: device not accepting address 3, error -110


This is the specific error from the logs. I have upgraded Intel machines remotely via yum and they are exhibiting the same result. The usb disk is no longer detected.

The bug tracker appears to be emailing me login details.

Is there an easy way to downgrade remotely to SME 7.3? 
I still have a 7.3 machine running and it does not have these issues.

[RBT]

Bug 5121


Is there an easy way to rollback to 7.3?

[Stefano]

Is there an easy way to rollback to 7.3?

no, at all

but you could try to boot with one of the previous kernel..

you should have the last 3.. at boot prompt, hit any key and choose

HTH
ciao
Stefano

[RBT]

I don't have a keyboard or screen  .... which kernel is the 7.3  one?

[CrayXMP]

I don't have a keyboard or screen

SSH to your box and edit /boot/grub/grub.conf and select the desired kernel with the "default=" setting.
It sets which listing to boot as default. 0 is the first, 1 the second etc.

Code: [Select]

#### 1st listing boots with default=0
title SME Server (2.6.9-78.0.13.EL)
####
#### 2nd listing boots with default=1
title SME Server (2.6.9-78.0.8.EL)
####
Hope it'll help !

[Stefano]

I don't have a keyboard or screen  .... which kernel is the 7.3  one?

Hi

here you'll find the kernel rpms used by Centos 4.6 (i.e SME 7.3)

if you know what are you doing you'd try to install an old kernel.. be aware that if you update your server, only 3 last versions of kernel will be installed, the other (the older) will be uninstalled.. so pay attention when you update your server.

HTH
Ciao
Stefano

P.S. to choose which kernel to use, follow CrayXMP advice

[RBT]

Thanks guys, got it sorted. swapped the external usb cases for e-sata cases