4400 All UA set Hange

phengvanna · Post by **phengvanna** » 08 May 2006 03:17

4400 R 5.0 Unx Multi CPU (Main/Standby)
All UA set hange, then give shutdown -i now to reset main CPU,
System reset then All UA set are work fine. After this
- Can not telnet to main CPU IP@
- Standby CPU and telnet and also it give status as Standby event we reset main CPU

I gonna crazy because of that
Any I dea please advise

Thanks
Vanna.

Post by **frank** » 08 May 2006 13:19

What if you get into the CPU from the console ?
Can you see the IP status ?

Any backtrace ? Any incidents ?

axel_h · Post by **axel_h** » 09 May 2006 15:57

For various reasons I recommend NOT to use the internal ethernet link between the CPUs. We had main instable main CPUs blocking all ethernet access to the slave CPU. If you change to two separate links, DO NOT forget to disable the internal link using jumpers. Which 5.0Ux release is it?

Post by **frank** » 09 May 2006 21:30

Axel,
Can you give more details ?

axel_h · Post by **axel_h** » 10 May 2006 01:25

Hi Frank, hi Vanna!

As for duplex CPUs: I only use separate ethernet-links to each CPU for some reasons. First, when a CPU has a problem and shuts down it is likely that this one blocks all ethernet access to the main CPU. Or when you have to shut one down by remote-access and tell the customer on site to pull this CPU out, you have to explain him to check that the network-link is put on the correct CPU. In a contact center we had an external CSTA-application which caused the csta_mono-process to crash an blocked the ethernet-link. In that case all CCS have no more access and you have to go on site and connect serial.
When you use two ethernet-links you MUST disable the internal ethernet-link, otherwise you cause a loop throughout the whole network. The customer would tell you at once! On CPU 5 and 6 you have to remove or add a jumper, it is in the documentation or check TC0403.
Also I had CPUs which could only be accessed serial and you I got: "CANNOT FORK, TRY AGAIN". In this case it is often not even possible to make a shutdown. In this case try removing PSAL or refer to TC0302. This one tells you to press CTRL G O. Warning: in this status of the CPU it often causes the OS to crash and reboot. Under Unix the file system sometimes crashes.

@Vanna: how did your problem turn out? Can you give any details? Have you seen incidents on the console?

OK guys, I have to leave office now.

Post by **frank** » 10 May 2006 09:34

Ok..
For the loop you are talking about, the customer HAS to turn on what's call SPANING TREE, which will tell to the switch where it's connected not to loop. I had an install with over 800 IP phones, and both ethernet connected. Without spanning tree first, they brought down the network. With spanning tree finaly, they got it up and redundant. I believe also that if you have only one ethernet connected, then when your cpu changes role, all the hybrid links and ip phone will go down. From what I remember, if you have dual ethernet, you can avoid that.

Regarding the "cannot fork" when you try to telnet, I saw that only on a 4.2 and never since. What release what this cpu you are talking about ?

Thx !

axel_h · Post by **axel_h** » 10 May 2006 11:37

Hi Frank!

Ok, itÂ´s a bit off topic now, but for spaning tree, I guy told me the same but if you disable the internal ethernet itÂ´s better at all...
As for switchover: if you manage lanpbxbuild with all IP-addresses it might help. As for hybrid-links, I guess it doesnÂ´t matter, cos all ABCAs go down.
About cannot fork: the last one I was was d23137m.

Post by **frank** » 10 May 2006 14:08

Ok.. I believe the FORK stuff is working now.
Regarding the IP phone down, I am pretty sure they will in all case, whatever way you manage the lanpbx.. If you have only one connection to the Ethernet, and if the main CPU reboot, it will disconnect for a second ..

If I have a chance to try 2 Ethernet with IP phones, I'll let you know..
Maybe tomorrow.