hp9000 Server Stalls/hangs and automatically reboots

[BLADE_1]

Dear All,

We have a hp9000 L1000 series Server running ux 11.0. Recently we had a ram upgrade adding 500megs to the existing 1.5gigs. The application (SAP) and Database (oracle) reside on the same server.

The problem started occuring since last week. The server telnet/ftp halts (we cannot telnet or ftp but the SAP application is working and we can also ping) or the System comes to a complete stops functioning and automatically reboots (we don't even get ping response during that time).

i checked for core file or system messages.
this is what i found.
1. shutdown log --> Reboot after panic - Data page fault

2.dmesg -> processor 0!
mpc_bindlwp: Overriding conflicting mandatory binding!
mpc_bindlwp: Migrating process 15002 from processor 1 to processor 0!

3. gsp report --> attached

[S.K. Chan]

There are numerous patches that are related to "mpc_bindlwp" error. From the patch DB in ITRC these are the ones that you need to pay attention to. Take a look at the description of each of them (look for keyword mpc_bindlwp) and I would recommend loading them to see if it helps.
PHKL_27770 (s700_800 11.00 vfs cumulative;resolve SG/vsar incompat)
PHKL_26001 (s700_800 11.00 diag2 must be MP-safe;no panic when logging)
PHKL_18543 (s700_800 11.00 PM/VM/UFS/async/scsi/io/DMAPI/JFS/perf)
PHSS_21922 (s700_800 11.00 STM, CPU & PCI Info tools fix unsafe MP call)

[Ravi_8]

Hi,

could you increase the swap by another 256MB

[Ravi_8]

Hi,

could you increase the swap by another 512MB

[BLADE_1]

hi,

I ran stm and checked the cpu log--
i also found this warning

WARNING: Current loading of DIMMs failed thermal loading order check. Operation with this DIMM loading could lead to unreliable system operation and degraded device lifetimes. Refer to DIMM loading rules and re-distribute DIMMs.

[barrett cooper]

You may have had a bad dimm installed or they were put in one or more wrong slots. use the following to check the order and config of the memory:

echo "selclass qualifier memory;info;wait;infolog" | /usr/sbin/cstm

Everything should be mapped out as you would expect. If not get ahold of HP to replace or move the memory to the correct location.