T
Tang Thim Sing
Hi
We have been tracking the source of this problem and we are pretty sure that
the deadlock was not caused by our application. Most interestingly, we have
also found from monitoring our web page requests from the log captured by
the Internet Services Manager (in WINNT/system32/logfiles) that; on a few
occasions when this error was reported in the event viewer, there were no
users using the system.
I'm also aware that this problem is also reported on the Microsoft support
site, being a false alarm due to slow write to the client. But in our case,
during those times when there were no users???
We are also up against the timeout problem due to connection pool maxed out.
We are seeing a pattern between these 2 problems. New connnection pools are
created each time the Aspnet_wp.exe is recycled. When this happens several
times, the number of connection pools goes up. The inactive pools would not
disappear until we restart the server. We noticed that there is a higher
chance of the timeout problem when the number of connection pools is high
(7-8).
We are also aware that the textbook answer to this timeout problem is to
ensure that the connections are explicitly closed. While we are tidying up
our system to ensure this, we have also experimented with the connection
lifetime value. By setting a very small value (5 minutes) in a controlled
environment, we found that all the connections in the pool are destroyed
after 5 minutes or so after we logged off our system, even after having
requested those pages that we know we have not explicitly closed the
connections. This implies that the connections are being returned to the
pool.
Regards
Tang Thim Sing
We have been tracking the source of this problem and we are pretty sure that
the deadlock was not caused by our application. Most interestingly, we have
also found from monitoring our web page requests from the log captured by
the Internet Services Manager (in WINNT/system32/logfiles) that; on a few
occasions when this error was reported in the event viewer, there were no
users using the system.
I'm also aware that this problem is also reported on the Microsoft support
site, being a false alarm due to slow write to the client. But in our case,
during those times when there were no users???
We are also up against the timeout problem due to connection pool maxed out.
We are seeing a pattern between these 2 problems. New connnection pools are
created each time the Aspnet_wp.exe is recycled. When this happens several
times, the number of connection pools goes up. The inactive pools would not
disappear until we restart the server. We noticed that there is a higher
chance of the timeout problem when the number of connection pools is high
(7-8).
We are also aware that the textbook answer to this timeout problem is to
ensure that the connections are explicitly closed. While we are tidying up
our system to ensure this, we have also experimented with the connection
lifetime value. By setting a very small value (5 minutes) in a controlled
environment, we found that all the connections in the pool are destroyed
after 5 minutes or so after we logged off our system, even after having
requested those pages that we know we have not explicitly closed the
connections. This implies that the connections are being returned to the
pool.
Regards
Tang Thim Sing