Gino Ruopolo wrote:
...
Updates after 2 days of testing:
DL585, Solaris Express 7/06
With the settings explained before system is up and running. No crashed EXCEPT
when using
SAN related commands. fcinfo, luxadm, cfgadm caused frequent panic.
DL585, S10u2
With the settings explained before is crashing under load.
Last occurrance:
...
Aug 27 14:56:24 fb2 unix: [ID 836849 kern.notice]
Aug 27 14:56:24 fb2 ^Mpanic[cpu0]/thread=fffffe8000d1ec80:
Aug 27 14:56:24 fb2 genunix: [ID 335743 kern.notice] BAD TRAP: type=e (#pf Page fault) rp=fffffe8000d1e7f0 addr=212cff68 occurr
ed in module "cpqci" due to an illegal access to a user address
Aug 27 14:56:24 fb2 unix: [ID 100000 kern.notice]
Aug 27 14:56:24 fb2 unix: [ID 839527 kern.notice] sched:
Aug 27 14:56:24 fb2 unix: [ID 753105 kern.notice] #pf Page fault
Aug 27 14:56:24 fb2 unix: [ID 532287 kern.notice] Bad kernel fault at
addr=0x212cff68
...
Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1e700
unix:die+da (80, 1, 500000000, fffffffff)
Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1e7e0
unix:trap+cb6 ()
Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1e7f0
unix:_cmntrap+11b ()
Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1e8f0
cpqci:cpqci_attach_state+303c44e5 ()
Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1e9b0
cpqci:CpqCiDrvCreateAdapter+5b5 ()
Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1ea70
cpqci:cpqci_attach+1da ()
Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1ead0
genunix:devi_attach+8f ()
Hi Gino,
the problm is in the "cpqci" module, which I assume is something
from HP/Compaq which you're using to manage your SAN.
This line is particularly worrying:
> Aug 27 14:56:24 fb2 genunix: [ID 655072 kern.notice] fffffe8000d1e8f0
cpqci:cpqci_attach_state+303c44e5 ()
notice that the function's offset is quite large? That's a sure sign
that there's a problem with that driver.
I don't think Sun supports the cpqci driver, so you need to contact
the vendor which does and get them to work out what is going wrong.
You might want to enable kmem_flags=0xf in /etc/system and try running
with that for a bit then try to reproduce the problem. That should
provide more information for the other vendor to work with.
cheers,
James C. McPherson
_______________________________________________
opensolaris-discuss mailing list
[email protected]
