Hi all,
We are experiencing a strange behaviour on our sun clients
in that they suddely decide to pick up the time from an IBM
server that unfortunately are running one hour ahead [date -u].
I do relize that this is a problem but it seems as if at some
point in time a clinet decides to get the time from this "bad"
clock. This server is only running as a fileserver, and everything
works just fine until a client decides to reset the internal
client clock to the "bad" time; the client chokes. The other databse/file-
servers all agree to the right time. So as long as the client
asks/updates its own clock from these three we're fine.
We have tried to inhibit the client from adjusting time accoring
to the "bad" server by utliizing the setserverprefs method in the client.
We are uncertain if this works or not.
My question: What scheme are used when a client are setting/adjusting
its internal clock? What makes it suddenly pick upp the time from a server
not even in the /usr/vice/etc/CellServDB file? Which servers are asked?
What's puzzling is the fact that anything at all works having this
fileserver ahead one hour with respect to the rest. Is the file
functionality/server less sensitive to having a bad clock?
The volumes/files on this server are accessable and all fine.
What would happen if we shutdown this server, set the clock back one hour
and the system will find datestamps "from the future" when it comes up
again? Maybe we should shutdown -s, set the time, leave it off for one
hour, *then* turn it back on?
Experience anyone?
...please don't ask me how this server got into this dead end street
in the first place...
Thanks,
/peo
