cancel
Showing results for 
Search instead for 
Did you mean: 

ML370 G3

 
D. Rick Anderson
Occasional Visitor

ML370 G3

We're having a problem with a couple of new ML370 G3s that we just purchased. Here's the configuration:

Dual Xeon 2.4GHz
1.5GB RAM
RedHat 8.0
Software RAID 5
Zope 2.6.0

Approximately every two days they completely lock up. No ping response, nothing on the console, no SSH access. We have to shut them down completely and bring them back up. I had thought it was some kind of hardware issue on the first one, until we got the second one up and it did the same thing. We're running this same config on several G2s and G1s using both software and hardware RAID, but have yet to run into this. Anybody know what this is?

TIA

Rick
9 REPLIES
Greg Carlson
Honored Contributor

Re: ML370 G3

Rick,

Try running with a Single CPU and see if it is related to the mutltikernal. Look at this post:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x642431ec5e34d711abdc0090277a778c,00.html

Here is the NOS compatability spreadsheet
ftp://ftp.compaq.com/pub/products/servers/os-support-matrix-310.pdf

RH8.0 is supported on the ML370 G3

Ciao,
Greg
Lets Roll!
Greg Carlson
Honored Contributor

Re: ML370 G3

Rick,

Also, are you at current levels on your BIOS and drivers? Make sure you are at current levels which are available here:

http://h18000.www1.hp.com/support/files/server/us/locate/87_1341.html

Ciao,
Greg
Lets Roll!
D. Rick Anderson
Occasional Visitor

Re: ML370 G3

Thanks for your reply. I located the BIOS update for the G3 and have installed it on both servers, so we'll see what happens. I hate problems like this because there is no way, other than watching it for days, to determine if you've fixed it.

The other thing I've noticed is that /proc/cpuinfo lists 4 cpus instead of just 2. I've looked and there are indeed only 2 physical processors, but that may have something to do with the Xeons. This is the first time I've used them.

Thanks again,

Rick
Greg Carlson
Honored Contributor

Re: ML370 G3

Rick,

4 CPU's showing with only two physical CPU's is due to Intel's Hyperthreading:

http://www.motherboards.org/articlesd/motherboard-reviews/1212_4.html

Ciao,
Greg


Lets Roll!
D. Rick Anderson
Occasional Visitor

Re: ML370 G3

Ok ... the servers both went down again today (feb. 18th). This time I disabled the hyperthreading to see if it will help.

I'll cross my fingers.

Rick
Greg Carlson
Honored Contributor

Re: ML370 G3

Rick,

Any update on the stability of your server?

Ciao,
Greg
Lets Roll!
Mario Obejas
Occasional Visitor

Re: ML370 G3

I'm following this thread because I'm experiencing the same thing.

2 x 2.8GHz processor
2.5 GB RAM
External U160 SCSI attached IDE RAID box
Proliant ML370G3,
Redhat 8.0 Professional.

Running the stock uniprocessor kernel (2.4.18-24.8), everything stays up and is happy. When I switch back to the smp kernel, I get a frozen system within 24 hours. Once it froze on me as I was typing in a text editor.

I call this a "hard freeze". In the frozen state, the system will not respond to a CTRL-ALT-DEL. It must be powered off.

I have tried disabling Hyper-threading to no avail - it still freezes. I am now (i.e., today, March 04) trying disabling the MPS table (Bios; Advanced Functions; MPS Table = Disabled).

This system is preproduction as a file server, NIS Master, Web Server, Print server (we're doing a lot of consolidation with it as the target). I have not had any problems running the SMP kernel on a Dell system, so I'm thinking it's Compaq specific.

There is no thermal issue either. The air is not even warm coming out the back. We are only running with two 18GB drives, and one SCSI PCI card.

SIG. Amedeo Martinacci
Occasional Visitor

Re: ML370 G3

I have the same problem. Any news about a possible solution ?
TIA

Amedeo
Michele_20
Occasional Visitor

Re: ML370 G3

Have you got some Intel NICs in those servers? I've ecnountered similar "random" problems with another (non-Compaq) server that had one Intel Pro100s. There seem to be some Intel NICs that don't do well with standard eepro100 kernel driver and hang this way after a random amount of time. They work properly with the e100 driver (the "alternate" driver in kernel configuration). I'm telling you this because in my recently bought ML370 the two optional Compaq NICs were Intel-based. My NICs are probably ok, but, since my previous experience, I think I'll run them with the e100 driver...
The other server ran fine for over two months since I've switched to the e100 driver, while it didn't stay up for over a week with the eepro100 driver, so I assume that was the solution. You could try and test it...
unless your coffee machine isn't connected to an ups, you are still ok!