off time sync issues in messages file

[iandid...@googlemail.com]

Solaris 10 10/09 s10x_u8wos_08a X86


A collesague noted the following in a /var/adm/messages file

...
Oct 13 03:20:00 tpa-etta root: [ID 702911 user.alert] this is for big brother
Oct 13 08:10:15 tpa-etta unix: [ID 836849 kern.notice]
Oct 13 08:19:33 tpa-etta ifconfig[28]: [ID 293258 user.warning] libsldap: Status
: 2 Mesg: Unable to load configuration '/var/ldap/ldap_client_file' ('').
Oct 13 08:10:15 tpa-etta ^Mpanic[cpu2]/thread=fffffe80000adc60:
Oct 13 08:20:09 tpa-etta nscd[217]: [ID 781775 user.error] libsldap: Can't disab
le Metaslot for self credential mode
Oct 13 08:19:33 tpa-etta last message repeated 10 times
Oct 13 08:10:15 tpa-etta genunix: [ID 335743 kern.notice] BAD TRAP: type=e (#pf
Page fault) rp=fffffe80000ad920 addr=0 occurred in module "bnx" due to a NULL po
inter dereference
Oct 13 08:19:33 tpa-etta ifconfig[39]: [ID 293258 user.warning] libsldap: Status
: 2 Mesg: Unable to load configuration '/var/ldap/ldap_client_file' ('').
Oct 13 08:10:15 tpa-etta unix: [ID 100000 kern.notice]
Oct 13 08:10:15 tpa-etta unix: [ID 839527 kern.notice] sched:
Oct 13 08:19:33 tpa-etta last message repeated 10 times
...

As you can see there are entries that seem to suggest something is out of sync when being logged.

Entries prior to 08:10/08:19 do not demonstrate this rather bizarre behaviour.


Given the rest of the timings in the messages log subsequently, it would appear that it is the 08:19 and 08:20 timestamps that are the ones out of sync.


FWIW the system uses ntp time sync and there are no obvious issues surrounding the box generally associated with any time syncing. The box is not a syslog server either.

Has any one any suggestions as to what may be happeninig here?

cheers

Ian

[hume.sp...@bofh.ca]

iandid...@googlemail.com wrote:
> Oct 13 08:10:15 tpa-etta genunix: [ID 335743 kern.notice] BAD TRAP: type=e (#pf
> Page fault) rp=fffffe80000ad920 addr=0 occurred in module "bnx" due to a NULL po
> inter dereference

Are you sure this box didn't crash and reboot?

--
Brandon Hume - hume -> BOFH.Ca, http://WWW.BOFH.Ca/

[iandid...@googlemail.com]

cheers brandon... yes it did. My colleague is investigating it... i am only periopherally involved (as I am the one with usenet expereince basically!).

Is that pertinent? (the fact that it rebooted, not that I have some USENET expereince)

cheers

ian

[cindy.sw...@gmail.com]

On Tuesday, October 16, 2012 7:33:11 AM UTC-6, (unknown) wrote:
> On Tuesday, October 16, 2012 1:35:48 PM UTC+1, (unknown) wrote:
>
> > iandid...@googlemail.com wrote:
>
> >
>
> > > Oct 13 08:10:15 tpa-etta genunix: [ID 335743 kern.notice] BAD TRAP: type=e (#pf
>
> >
>
> > > Page fault) rp=fffffe80000ad920 addr=0 occurred in module "bnx" due to a NULL po
>
> >
>
> > > inter dereference
>
> >
>
> >
>
> >
>
> > Are you sure this box didn't crash and reboot?
>
> >
>
> >
>
> >
>
> > --
>
> >
>
> > Brandon Hume - hume -> BOFH.Ca, http://WWW.BOFH.Ca/
>
>
>
> cheers brandon... yes it did. My colleague is investigating it... i am only periopherally involved (as I am the one with usenet expereince basically!).
>

Hi Ian,

Do you have the crash dumps ... Can you provide the panic stack?

Thanks, Cindy

[iandid...@googlemail.com]

Hi Cindy,

my colleague has /var/crash/<box name>/vmcore with a date and timestamp of about the correct time.

what do i need to do to garner the panic stack from that file?

cheers

ian

[cindy.sw...@gmail.com]

[cindy.sw...@gmail.com]

Hi Ian,

The steps are something like this if vmdump.0 or
substitute the correct #.

# cd /var/crash/sys-name
# savecore -f vmdump.0
# mdb -k 0
> :: stack
> $q

Thanks, Cindy

[iandid...@googlemail.com]

thanks!


> ::stack
bnx_xmit_ring_xmit_mblk+0x8e()
bnx_m_tx+0x69()
dls_tx+0xe()
dld_wsrv+0x12d()
runservice+0x5a()
queue_service+0x3e()
stream_service+0x63()
taskq_d_thread+0x1ba()
thread_start+8()
> $q

cheers

Ian

[cindy.sw...@gmail.com]

Ian,

I see CR 6874776 with a similar stack but it is fixed in s10u9.
Do you see bnx errors (link up/link down) in /var/adm/messages as well?

My view is not as vast as someone in support so you might consider filing
a service request.

Thanks, Cindy