SQL Server 2005 SP2 CU2 64-bit Ent Edition, Active/passive Cluster. Involved in transactional replication both as a subscriber (to about 350 publishers), and a publisher (to 1 subscriber). Distribution runs on seperate servers.
both nodes are Dell 6850'x 8 cores, 16ÃÂ BG ram, running on a CX3-80 SAN
24x7 sales website - this is the OLTP DB supporting it.
CPU typically around 10%-20%. Still gettingÃÂ specific about CPU during the actual failures, but it doesn't spike, and no MOM alerts fired.
User Connections: historically was around 500-600, but during this week, some changes made to on the distribution server handling the the 400 publications the server subcribes to, has been changed, and we not have around 250 user connections.
We have had quite a few issues this week - but none before this week.
Other than a change to the way we are getting the transactional replication data into this server (reducing the number of publishers connecting to the OLTP DB), no *known* system changes - no SQL upgrades, no OS upgrades, no hardware changes/ugrades, no application/DB releases. We have a semi-decent change control system, so while we will chase those areas to ensure nothing unknown happened, they aren't our primaryÃÂ area of suspicion.
What we are seeing is a large amount (several hundred) of the following messages (taken
View Complete Post