Has this problem ever been seen outside of one particular IOC? I can’t help but notice that many things are misbehaving other than this assert. For example, eight of the server’s input threads are either CPU starved or unable to get a lock that they need. I also see “callbackRequest: cbHigh ring buffer full” which might happen if device support is posting IO interrupt events at a high rate.
Has this problem ever been seen outside of one particular IOC? I can’t help but notice that many things are misbehaving other than this assert. For example, eight of the server’s input threads are either CPU starved or unable to get a lock that they need. I also see “callbackRequest: cbHigh ring buffer full” which might happen if device support is posting IO interrupt events at a high rate.
edited on: 2009-07-01 17:09