Have a problem that I haven't experienced at any customer.
Setup:
Microsoft Cluster Server on 2 Windows 2003 Advanced Servers
PSQL V8... last SP
Been operational for over a year.... been very stable.
2 weeks ago... then again twice today.. the following scenario had
occured:
Mid production day - reasonable volume.. .nothing overly "heavy"...
suddenly ALL clients are disconnected from the server. "Your
connection to the server is no longer working... etc"
I log into the Cluster... ie.. the Primary/active Node... via Remote
Desktop
I fire up PCC to check the config of the engine... the "Server"
config menu is not even there! Only the "Client" menu options.
I open PVSW.LOG... the first entry I see for today's date is:
W3CACHE.. W3DBSMGR.EXE... "Resourced Allocated"
I simply stopped/started the PSQL services.. and they came back up...
as a result.. in PCC.. the "Server" menu is now displayed as it should
be. In addition, in PVSW.LOG... I see...
W3CACHE... W3DBSMGR.EXE... "Resources Released"... followed by
NTMKDE..... NTDBSMGR.EXE... 'Resources Allocated"... etc... then a
number of rollbacks to tables, etc.
This happened twice today... and once 2 weeks ago... prior to that..
never. And only at this customer.
Doesn't the W3DBSMGR.EXE reference the WORKGROUP ENGINE??
There are NO events in the Windows event log to indicate any problem
with the cluster itself. No failover had occured to node 2... node 1
stayed operational the entire time.
Has anyone ever seen this before?
John
