[OpenAFS] HELP! We 've lost our sync site

[James M Mosley]

All,
	We need immediate help!  We have been unable to establish a sync
site for about 6 hours.  All 3 of our database servers are up and appear
to be perfroming the election as expected.  However, the server that
should be come the synce site doesn't.    Here is some output from udebug
on that server:

as-sm1# udebug as-sm1 7002 -long
Host's addresses are: 152.15.10.70
Host's 152.15.10.70 time is Sat Jan 10 13:59:36 2004
Local time is Sat Jan 10 13:59:39 2004 (time differential 3 secs)
Last yes vote for 152.15.10.70 was 3 secs ago (not sync site);
Last vote started 3 secs ago (at Sat Jan 10 13:59:36 2004)
Local db version is 1073480540.254
I am not sync site
Lowest host 152.15.10.70 was set 3 secs ago
Sync host 0.0.0.0 was set 1073761176 secs ago
Sync site's db version is 1073480540.254
0 locked pages, 0 of them for write

Server (152.15.13.7): (db 0.0)
    last vote rcvd 5 secs ago (at Sat Jan 10 13:59:34 2004),
    last beacon sent 3 secs ago (at Sat Jan 10 13:59:36 2004), last vote was yes
    dbcurrent=0, up=1 beaconSince=1

Server (152.15.30.27): (db 0.0)
    last vote rcvd 4 secs ago (at Sat Jan 10 13:59:35 2004),
    last beacon sent 3 secs ago (at Sat Jan 10 13:59:36 2004), last vote was yes
    dbcurrent=0, up=1 beaconSince=1
as-sm1#


The only strange thing we have noticed is that when we attempted to
stop/restart the database servers to see if the condition we clear itself
up we saw as-sm1 become the sync site (as it should) but it claimed it was
a sync site for a negative number of seconds.  The amount of time seemed
to refer back to about the time we started seeing the problem as evidenced
by the last time the local database files were updated.

All three database servers our running Solaris 9 and OpenAFS 1.2.10.

We need help soon.  Thanks.

Mike

-------------------------------------
Mike Mosley  				Email: jmmosley@uncc.edu
Systems Software Developer 		Phone: (704) 687-3522
College of Engineering, UNC-Charlotte 	Fax: (704) 687-2352