Describe the bug
Hi, we are suffering high issues since we migrate to the last version 1.5, we have an institution with that schedule arround 200 sessions for the same range of time, and we have pool of 22 servers to attend this load, with the version 1.4 the load were distributed more equality, but in this new version, the Scalelite sends high number of meetings to the same node at the same time, instead of roundrobin to different servers.
Seems that it perform the same balacing algoritm for all the meetings that are entering at the same time, wihout taking account that he is actulally inserting already other meetings at the same second to the same node.
To Reproduce
Schedule over 100 meetings at the same time, and start them in very short time through the scalelite balancer, the Scalelite will send high number of meetings to the same node that have the lowest load, wihtout taking care that will overload it because he is inserting other meetings to the same server at the same time.
Expected behavior
The expected behaivor is that the scalelite spread the load arround all the lowest servers and have a way to calculate o measure the number of meetings that it should insert to the same node in range of time, example every 15 seconds no more than 5 meetings. Also a paramter to limit the number of meetings that he can start by server should help to reduce this overload issues.
Describe the bug Hi, we are suffering high issues since we migrate to the last version 1.5, we have an institution with that schedule arround 200 sessions for the same range of time, and we have pool of 22 servers to attend this load, with the version 1.4 the load were distributed more equality, but in this new version, the Scalelite sends high number of meetings to the same node at the same time, instead of roundrobin to different servers.
Seems that it perform the same balacing algoritm for all the meetings that are entering at the same time, wihout taking account that he is actulally inserting already other meetings at the same second to the same node.
To Reproduce
Schedule over 100 meetings at the same time, and start them in very short time through the scalelite balancer, the Scalelite will send high number of meetings to the same node that have the lowest load, wihtout taking care that will overload it because he is inserting other meetings to the same server at the same time.
Expected behavior The expected behaivor is that the scalelite spread the load arround all the lowest servers and have a way to calculate o measure the number of meetings that it should insert to the same node in range of time, example every 15 seconds no more than 5 meetings. Also a paramter to limit the number of meetings that he can start by server should help to reduce this overload issues.
Additional context