We have a periodic event where lots of connections hang on TM REQUEST (from
sys.dm_os_waiting_tasks) and make no progress.
KILL on the affected
session_id does nothing
The server runs database mirroring (2008R2). Pausing and resuming mirroring fixes the issue when it happens.
The session stuck in this state are unkill'able. The load on the mirror is the usual, not bursty. Once a high number of connections are stuck in this state, the system freezes up and the mirror stops replicating (all queries get blocked).
Anyone seen this before and know what is causing it? There is nothing in the error logs.