Hi,

We had a network/hardware issue with one of the replica set primary (hosted with AWS), and that caused the primary to switch to one of the secondary servers. However, our application went down, cause the driver did NOT update its cluster and was throwing this error

"No server is available matching preference: #"prod", :max_pool_size=>200, :connect=>:replica_set, :write=>{:w=>1}}, @server_selection_timeout=30>"

We use the following Mongo::Client options on the apps:
max_pool_size: 200
connect: replica_set
write:

hosts: ['prod-xxx-db4:27017', 'prod-xxx-db6:27017', 'prod-xxx-db7:27017', 'prod-xxx-db8:27017']

Environment:
Mongo Server 2.4
Mongo Ruby Driver 2.0.6
JRuby 1.7.20 (1.9 mode)

Event Timeline:
Aug 28 11:33 AM : The original primary (prod-xxx-db7) went inaccessible (not able to connect to it)
Aug 28 11:34-11:36 AM : Mongo Connection timeout errors on App servers
Aug 28 11:36 AM: prod-xxx-db6 was elected as primary
Aug 28 11:36-11:55 AM: Mongo connection errors ("no server is available") on app servers for queries to the 'prod' replica set
Aug 28 11:52 AM - App server restart fixed the errors.

Is there any known issue which causes this behavior? If not, please review and let us know if you need more info. Thanks!