for one of our customers, we have a TMG (7.0.9193.575) array with about 200 webpublising rules and a couple of server publishing rules. The TMG’s and EMS are running on different virtual machines.
The problem is, if I apply a change it takes a very long time one of the TMG is in sync with the EMS. Sometimes it will take 3 hours (during working hours). The other TMG is synced within 15 min. In front of the TMG’s there are hardware balancers and the load is equally.
The EMS is also used for other TMG arrays, without any problems.
I tried to copy a large file from EMS to both TMG’s, the result was no problem, so in my opinion there is no network problem.
To pinpoint, I created some connections verifiers to some web publishing sites. In the alerts, I see a lot of errors and warnings about the connection verifiers, mostly from the TMG with the slow sync. So , somehow there must be some relation. Also the connection verifiers for the webfarms, shows slow connections from the TMG with sync problems.
To simplify the webfarm connections verifiers, I changed them from TCP check to ping.
To lower the CPU load (between 15-30% during working hours), we changed the logging from SQL to text. No result.
We have a lot of TMG arrays, some with even a larger rulebase, however this one has the most web publishing rules. So there must be a relation with the syncing time and the number of web publishing rules.
Can someone give me a hint to resolve this problem ?