On 08 Aug 2007, Ulrich.Windl@rz.uni-regensburg.de wrote:

> Hi,
>
> in SCSI error messages in syslog HP-UX 11.11 reports an "lbolt"
> value. Can anybody explain what that is?
>
> Occasionally our system (that is up since September 2006) reacts a
> bit odd, and I noticed that the "lbolt" values are negative (usually
> they are positive). Could that couse some problem (when the system
> is up for a very long time)? To be concrete: a DDS tape in a DDS
> drive caused syslog to report SCSI Parity errors and SCSI
> resets. IMHO a SCSI parity error is a transport problem between the
> SCSI device and the HBA, and should not be caused by a bad tape in a
> drive (unless the reported parity error is nonsense).
>
> The latest installed SCSI-related pathes are
> PHKL_33425 (tape)
> PHKL_32090 (HBA)
>
> Regards,
> Ulrich

Hi Ulrich,

"lbolt" is an alias for a kernel global which contains the number of
clock ticks since the machine was booted. It's an unsigned long, so it
never really goes negative, although it does wrap.

The SCSI driver syslog messages print the value of lbolt with a '%d'
format directive, so it will print a negative value when the high bit
is lit. This is unrelated to the underlying cause of the messages.

Regards,
Carl Davidson

carl.davidson@hp.com writes:

> On 08 Aug 2007, Ulrich.Windl@rz.uni-regensburg.de wrote:
>
> > Hi,
> >
> > in SCSI error messages in syslog HP-UX 11.11 reports an "lbolt"
> > value. Can anybody explain what that is?
[...]
> Hi Ulrich,
>
> "lbolt" is an alias for a kernel global which contains the number of
> clock ticks since the machine was booted. It's an unsigned long, so it
> never really goes negative, although it does wrap.
>
> The SCSI driver syslog messages print the value of lbolt with a '%d'
> format directive, so it will print a negative value when the high bit
> is lit. This is unrelated to the underlying cause of the messages.
>
> Regards,
> Carl Davidson

Hi Carl,

thanks for the explanation. I had suspected that it's something like a "SCSI
transaction counter" to correlate kernel messages.
As a side note: It would be good if all SCSI-related kernel messages had the
"SCSI:" prefix. So I could "grep SCSI: /var/adm/syslog/syslog.log".

Regards,
Ulrich