We run fully duplexed. Our PG's each are linked to one of two chanels on the MAPD's. Sometimes, for no reason that I can find, one of the client connections will drop. This leaves the PG with only one active connection and seems to be related to some of the post route abnormal route end errors that we see here. Do you know what might be causing this connection to drop? I have verified that the hosts are enabled and the proper channel checkboxes are checked on the PG setup... This is on a Definity PBX.
You could put a sniffer and check on the connection where the drop is coming from, most likely it will be the ACD side.
Both ICM and the Definity are checking for a reliable network connection, if they feel that one of those link is congested they will drop it to reopen the socket once again.
Avaya should be able to drive you to how to actually verify when they are dropping it, sometimes it might be because to many requests are polling the link, by example.
Can you list out what versions of ICM, PBX, Map-D, etc. We've in the midst of diagnosing a similar problem we're having with a simplexed version 2.5 PG, Avaya CMS 6.
Avaya has the ability to increase the number of active associations on each link, so if you are running 2048/link, you could double one port to 4096 and just run on one. Alternately, you could configure your PIM to require two active links at a minimum, then if you lost one, you would have a failover and hopefully come up with two on the other side. I would pursue option 1 if you see this condition frequently.
Hello Al, you must comprove the sw version of the Definity, if you are migrated to version 8 you need a new dll installed in the PG.
If you don't have this dll, the communications PG -MAPD will be down (review de Cisco Documentation)
Before that, you also comprove the configuration of both PG's, must be the same or the duplexed running it's not possible.
Sorry, my english is very bad!!!
Just a quick check - is the network between your PGs and the MAPD connection 10Mb *half* duplex? Avaya I believe is still very strict on the requirement for the MAPD network connection.
Normally not having this set correctly can cause more serious connectivity problems and even problems at the software/firmware level on the MAPD card itself, but it's worth checking.
We had similar problems related to CMS. After going back and forth with Cisco and AVAYA, AVAYA came back and stated that the port speed should be set to 100MB/Full Duplex. CMS is set to 'auto-sense' but it does not seem to work 100%, so we are going to hard code it. Also check your swith port and PG and make sure they are 100/Full as well.
You may also want to get onto the latest hotfix. I checked quickly, and it looks like 4.1.5, HF 173 & 174 may apply to your problem.
We had the same issue. Cisco's solution for us was to set the PG's to 10/half. They indicated that the MAPD cards we had could not handle 100/full.
4096 Active Associations and have the NIC's set to 10/HALF. I gurantee that this will fix your issues. Also, if you are at 2048 Active Associations, it's about a $15,000.00 upgrade for the 4096. W went through the exact same thing here at Allstate and it has been smooth ever since.
Did you get to 4096 active associations by having it hard coded or by upgrading the CVLAN version? We went with multiple ASAI links (4 per side) because by our calculations some of our sites have about 4400 active association/dynamic associations. But we left the NICs at 100/full. Don't get the "max association" error anymore, but occasionally get the disconnect.
The PG's must be set to 10\HALF since the Avaya can not communicate any faster. We had the same issue with the NIC's being 100\FULL. Once we made the change, everyone was happy. If you have any questions, feel free to contact me at (847) 402-7368 or email@example.com