Prolian ML350 G3 hangs (debian & kernel 2.6.8)

[Piotr Dziubanii]

Hello,

we've got Proliant ML350G3 2xXeon 3GHz, 1.5GB RAM, Compaq Smart Array & raid1, Intel 2ports GB NIC
we're running debian sarge, kernel 2.6.8.1, hpasm ported from RH

Server hangs after couple of hours of heavy network load

we simulate heavy network load on Fast Ethernet swtched or directly connected network with linux iperf utility, we generate 80-90Mbps bi-directional traffic on 1st port of Intel NIC and integrated Broadcom GE NIC

after couple of hours server hangs, no debug messages, no kernel panic...
after 12min of hang time proliant automaticly restarts

[CA859951 · Accepted solution]

What bios revision are you ruinning?

There were some microcode fixes a while back to help in this regard:
http://h18000.www1.hp.com/support/files/server/us/download/21272.html

Updated to integrate the latest Intel processor support code into the System ROM. This works around an issue with the Intel Xeon processor that can result in erratic system behavior after prolonged usage. The failures were identified in a small percentage of Intel Xeon processors during simulated long-term reliability testing. These issues have only been seen in Intel's simulated environment, and no field issues have been identified. While the nature of the failure for systems experiencing this issue is difficult to quantify, systems may experience systems hangs, Windows blue screen, or Linux kernel panics. HP recommends customers transition to this new BIOS version to prevent the aforementioned issues.

Let me know if this clears up for you

-john

[Piotr Dziubanii]

we've got D14 02/24/2004

upgrade done, now we'll burn it and wait :)

[CA1192465]

Hello-
We have a ML350 G3, 2xXeon 2.2Ghz, 1GB RAM, Compaq Smart Array & Raid 1, gigE NIC

We're running debian sarge, trying to use kernel 2.6.8-1 SMP, but upon bootup, we recive a kernel panic error. A bit of research indicated it might be missing the aic7xxx driver in initrd, but that has been checked. Also, the unit has been updated to the latest bios available.

Thanks,
~Will

[Piotr Dziubanii]

hello!

after bios upgrade, there is no hang problem, for now

but proliant doesn't works fine and stable

when we start to generate network traffic, load increases to 3-5 and after few minutes NICs stops for few second
in linux logs we can see "ethX transmit timmed out, resetting"
on cisco switch we see "line protocol down" and "up" after few seconds

after upgrade, we observe strange cpu usage, OS
sees 4 cpus, ID# 0-3, CPU1 has 0% load and it looks like it doesn't work at all, ksoftirqd_cpu1 shows 100% cpu time all the time

referring to linux /proc/interrupts informations, CPU1 handles all irq for NICs

is it broken Xeon?

[Piotr Dziubanii]

Will, if You want to use Compaq Smart Array, You have to compile in Compaq Smart Array Support (ciss)

Device Drivers -> Block Devices -> Compaq Smart Array Support

or in .config:

# Block devices
#
CONFIG_BLK_DEV_FD=y
# CONFIG_PARIDE is not set
CONFIG_BLK_CPQ_CISS_DA=y
# CONFIG_CISS_SCSI_TAPE is not set
# CONFIG_BLK_DEV_DAC960 is not set
# CONFIG_BLK_DEV_UMEM is not set
CONFIG_BLK_DEV_LOOP=m
# CONFIG_BLK_DEV_CRYPTOLOOP is not set
CONFIG_BLK_DEV_NBD=m
# CONFIG_BLK_DEV_SX8 is not set
# CONFIG_BLK_DEV_RAM is not set
CONFIG_LBD=y

Best regards,

Piotr.

[BR109943]

Hello Piotr,

can you help me! we've got Proliant DL380 G3 and we're running debian sarge, kernel 2.6.8.1.
How did you ported HPASM from RH?

Thanks in advance,

Paolo

[Piotr Dziubanii]

Paolo: i did it with 'alien' utility.

greetz,

p.