|
[
Permlink
| « Hide
]
Eric Mueller added a comment - 19/Sep/05 05:33 AM
For clustered environments only a certain percentage of the available threads in the thread pool could be associated with triggers, so that the triggers can be distributed better amongst the nodes.
Unfortunately we are deciding not to put this into the 1.6 release.
Jasper and I have wasted basically 4 calendar weeks (and written novels worth of emails) pondering how to make this work. It is far more difficult than it seems on the surface - particularly when complicated by the new prioritization of triggers feature. In a nutshell, the goal of this enhancement was to improve performance of acquiring triggers, by reducing the number of queries in the case that more than one trigger is ready to fire. However, there is no way (that we have been able to discover) to implement this where it does not add an additional query to every trigger acquisition (even those most common times in which there is one (or zero) triggers ready to fire) -- so, the positive effects are basically more than negated, except in the less common case of there being many many of triggers ready to fire. In light of this, and the fact it would add a lot of complexity to the code, we have decided to abandon the idea for now - though the basic idea of acquiring more than one trigger at a time is very appealing. Backed out placeholder methods to return more than one trigger at a time.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||