SQL & network gurus: MS DTC problem

SQL & network gurus: MS DTC problem

Post by Luke Dougla » Sat, 18 Aug 2001 00:30:38



I have a number of remote SQL2K servers that use DTC to
call an sproc on my local SQL2K server (i.e. they call it
within a DISTRIBUTED TRAN and commit or rollback depending
on the error code). I'm primarily testing DTC at the
moment, so I have the remote servers executing queries
that execute distributed transactions as quickly as
possible.

Everything proceeds well at first. However, after the
remote servers have executed about 5000 distributed
transactions, DTC begins to fail -- the remote servers
receive error 8522 (Distributed transaction aborted by
MSDTC) and, if things are really bad, 911 (Communication
link failure). The local server's application event log
also fills up with event 17055 (Distributed transaction
aborted by MSDTC).

When I look at this in Performance Monitor, I see that the
average DTC response time hovers at 75 ms for a while,
then rises and eventually caps out at about 400 ms. As it
rises, the number of aborted transactions increases
(eventually rising to about 5/sec when the remote servers
are initiating about 25/sec).

At first glance, it's as if DTC is becoming saturated --
but if so, I'm not sure how to increase a timeout interval
or something to get it to queue transactions rather than
dropping them, or to execute some kind of internal flush.

Anyone else have a better guess (or, even better, a
solution)?

 
 
 

1. SQL & network gurus: MS DTC problem

I have a number of remote SQL2K servers that use DTC to
call an sproc on my local SQL2K server (i.e. they call it
within a DISTRIBUTED TRAN and commit or rollback depending
on the error code). I'm primarily testing DTC at the
moment, so I have the remote servers executing queries
that execute distributed transactions as quickly as
possible.

Everything proceeds well at first. However, after the
remote servers have executed about 5000 distributed
transactions, DTC begins to fail -- the remote servers
receive error 8522 (Distributed transaction aborted by
MSDTC) and, if things are really bad, 911 (Communication
link failure). The local server's application event log
also fills up with event 17055 (Distributed transaction
aborted by MSDTC).

When I look at this in Performance Monitor, I see that the
average DTC response time hovers at 75 ms for a while,
then rises and eventually caps out at about 400 ms. As it
rises, the number of aborted transactions increases
(eventually rising to about 5/sec when the remote servers
are initiating about 25/sec).

At first glance, it's as if DTC is becoming saturated --
but if so, I'm not sure how to increase a timeout interval
or something to get it to queue transactions rather than
dropping them, or to execute some kind of internal flush.

Anyone else have a better guess (or, even better, a
solution)?

2. How much do I need?

3. Linked Server & MS DTC

4. CMMS & Sybase SQL Anywhere

5. PROXY & MS SQL Server Client Network Utility

6. database tables report to word document????

7. 2nd try: SQL server network access restrictions, pls SQL gurus comment it

8. Sybase Performance

9. ms SQL & Website without Network socket

10. MS DTC Problem

11. Problems With MS DTC Service

12. Problems Installing MS DTC

13. MS DTC PROBLEM (URGENT)