Well, we performed the Adobe Connect upgrade to version 7 on Wed, July 30th. We had, previously to that, installed the upgrade on two test environments and both turned out fine.
There were no problems with the upgrade right away, it seemed to work fine, just a little slow, and the failover was not working properly.
Yesterday we had some serious issues with speed, people not connecting properly, html tags showing up on the pages sporatically....it was pretty much a mess.
Logging into one of our connect app cluster nodes, I found that a FMScore process was taking up 25% of the CPU continuously. Considering we have a quad core processor in there, it was flooding one processor. I killed the process, and I believe in the end, that's what fixed some of our problems. Only thing is, I had to reboot it to clear the stack, so the FMS would re-initialize and the problem would go away.
Anyway, I called Adobe with our problem and they were helpful. They solved the speed problem of our site, and let us know of an upgrade that might fix some of our other problems. We plan to install the upgrade tonight and see how everything goes.
Andy
Connect 7 Clustering
While running Connect 7 in a cluster, I believe doing data mirroring between the two nodes, have you noticed that the space consumed under the /content/7/ directories of connect has grown rapidly? Specifically it would be under a meeting or a few meetings and only on one node of the cluster possibly under a /input/ directory for that meeting with one or more .flv files approximate 212,993KB in size. We just recently began running a Connect 7 cluster and on multiple occasions we've seen this happening, wondering if it's happening only in cluster environments, because we haven't seen it on single node instances. Additionally, on the other node we see small size /7//output/ directory with smaller .flv files and meta data xml files.
There seems to be a few other problems with Connect 7, we need to upgrade to SP1 to resolve another issue and are in the process of testing the release.