Shelves dropping on Bascul

[MartinB]

Hi Guys

I have a complicated error I need some assistance on please.

I have a network with 8 Nodes and a lot of remote shelves from each node. Complete Alcatel network.
All Nodes have redundancy with 2 IBM appliance servers as XA and XB. (001001-001008)

On the "Main node (001001) it has happened now for the third time where some digital extensions on some of the remote shelves lose speech completely.

When I run the command "represent d <extnr>" I get different INTIP3 cards and in some cases INTOF cards responsible for carrying the speech so I cannot pinpoint a specific card as being faulty.

The only way to quickly solve the issue as users are complaining is to do a bascul BUT this causes a different problem - The 6 shelves belonging to Domain 20 all drop and in order to get them up and alive again is to wait for the TWIN command to show all ready and do ANOTHER bascul. It does not matter if I am on XA or XB when the problem occurs, I need to do a "double" bascul to get the cards up. So its not config error on a specific CPU.

If I set the pabx to switch CPU's at night with bascul or reboot of the main cpu, the shelves eventually come up on the other cpu but I have not tested the time duration. I only see the next morning all is okay. I will test this tonight.

The loss of speech issue is also not only on those 6 shelves, however the 6 shelves are the main shelves where all the intof cards and IP cards point to the other shelves.

If I do a represent and I see that for instance card 4-22 (INTIP3A) is responsible for carrying the voice for a digital phone on shelf 4-20 for example then I check with ippstat which IP phones are connected to that board and dial the IP phone to test if there is voice...and there is voice...it seems like only digital phones losing speech.

What can cause the shelves to not stay in service on Bascul? They drop a couple of seconds after the bascul was done.
I have done mastercopy so both CPU's are identical.
There are more than enough compressors on all the INTIP3 cards.
There is no compression on the domains.

I suspect a data config issue but cannot prove it.

hope this makes sense
thank you
Martin

[tot3nkopf]

Start with simple things and check your IP configuration for al INTIP/GD/GAs and CPU configuration. If that is ok investigate further and see if you have ports blocked (firewall for example)....
Anyways we cannot guess.. we do not know your configuration, your data network topology...

[MartinB]

Hi tot3nkop. I have checked all the intip cards and they are fine with regards to programming . Lanpbx is correct. It is as though the xm ip address is not releasing the mac address of the cpu when bascul gets done. It happens from both cpu's so the intip cards are definitely correct or this would have been isolated to xa or xb. I only recently took over this site and is trying to sort out all the issues they have been struggling with.
The cpu's are both in non ip domain 0 and all the intip cards from shelf 1 - 6 (20 x intip3 and 2 x ioip) are in ip domain 20. Same ip range as the cpu's and all on the same vlan. Nothing on firewall blocking. Both the cpu's are plugged directly into the core 9800 switch.

I will keep looking for the problem on all the progranming just thought i would ask here if there is anything obvious I am missing that someone can pointout to me.

Thanx
Martin

Sent from my SM-G920F using Tapatalk