Hi,

I have an interesting question. We use a system monitoring tool called
SMC from SUN/Halcyon that monitors the db production servers. It is
showing a -1 in the running column for processes that are running
long.

Anyway, Halcyon pointed out an issue with Sybase and o/s timeclocks
and synchronization.

I checked all EBFs etc and we are running a version that this has been
corrected in....we use ASE 12.5.0.3 EBF9072 ESD#1 applied. It runs on
Solaris O/S version 9.

If you issue select getdate(), and run date command in Unix, the times
match. However, if you issue the date command almost simultaneous on
the command line and then select the spid from sysprocesses table for
your established isql session connection, the loggedindatetime column
shows a 2 minute difference in time.

Does anyone experience this?

Sybase is running two minutes faster than the o/s system time. We use
o/s Network Time Protocol and Sybase has been recycled quite
regularly, even as far as just being recycled on MOnday.

Appreciate any feedback/info.

Thanks
Cynthia

clangev (AT) yahoo (DOT) com (Cynthia) wrote in message news:<da0df555.0402260735.47f055eb (AT) posting (DOT) google.com>...

Sure, this is possible. The getdate() function makes a system call
to the OS to get the most accurate current date and time, but
for many other functions (crdates for objects, loggedindatetime,
begin and commit times in log records, etc.) ASE uses an internal
clock - this is set at boot time, and then once a minute ASE
checks the OS clock and either speeds up or slows down the internal
clock to try to stay in synch (note that the internal clock is
never run backwards though). It can get out of synch, though usually
not by much. ASE does depend on the regular receipt of SIGALRM
from the OS for the running of this clock - values very much out
of synch are a classic symptom of the missing SIGALRM issue on
Solaris - see http://www.sybase.com/detail?id=1016173 for details.

-bret

bret (AT) sybase (DOT) com (Bret Halford) wrote in message news:<34c88f2b.0402261600.7c5c860a (AT) posting (DOT) google.com>...

Thank you very much Bret for your answer. We have tried the Unix
commands from the above article to check the system and we also do not
experience any of the other symptoms listed as indicators for missing
SIGALRM. So, I guess the internal clock to Sybase is running faster
than the os as we pretty consistently get a 2 min difference between
the Sybase internal clock and the os time.

Thanks again for your help.

Cynthia