Red Hat Bugzilla – Bug 83592
(NET TG3)SuperMicro P4DL6 locks up on network data transfers
Last modified: 2013-07-02 22:09:54 EDT
Description of problem:
I have a SuperMicro P4DL6 dual-Xeon server with RedHat 7.3 that is locking up
on me, while doing network data transfer tests (on either the gigabit or
100baseT interfaces). At the suggestion of someone who encountered similar
probs, I installed an updated RedHat smp kernel 2.4.18-19.7.x,
appended "noapic" to lilo, and updated the AMI BIOS to Rev. 1.1, using the
following settings in the new BIOS (after loading the optimal defaults):
- Mem Timing -> Auto
- MPS 1.4 Support -> Enabled
- Hyperthreading -> Enabled
- Watchdog -> Disabled
- PnP O/S -> No
- Reset Config Data -> Yes
- Allocate IRQ to PCI VGA -> No
- PCI IDE Busmastering -> Disabled
- ACPI Aware O/S -> No
- Power Management -> Disabled
but I am still experiencing lock-ups, when doing file transfers b/t two of
these identical servers.
Version-Release number of selected component (if applicable):
2.4.18-3smp and 2.4.18-19.7.xsmp
Steps to Reproduce:
1. network two p4dl6 servers together
2. continuously stream data over a random unused port from one machine to the
other (i am using port 2000).
3. wait. before my kernel/bios updates, it took about a week to lock up, and
after the updates, almost instantly.
One of the two machines always locks up, and never the other one. sometimes,
it is the receiving server that locks up, and sometimes it is the one
initiating the data transfer.
no locking up!
I expect this is a motherboard/serverworks chipset issue, but supermicro has no
solutions, apart from the latest BIOS update, which i've already installed. i
am hoping others have experienced this w/7.3 and have come up w/workarounds.
Hopefully resolved in current errata. Can you confirm it is ok for you now ?
Yes, fixed. I believe it was RH kernel 2.4.18-24 that did the trick for me.
You can squish this bug.
Thanks for following up,