Thursday, November 7, 2013

IBM Sametime 9 - Media Manager components required startup order

EDIT:
I think this is fixed in the latest Media Manager Fix pack. HotFix1.

Today I discovered that there might be a specific startup order for the Media Manager servers for the Sametime 9 environment.

I had no Video conferencing option after a reboot, and the SP+CF servers SystemOut.log told me that a connection to the VMGR server was "denied".

So, I then shut down the SP+CF server, shut down the VMGR and of course the SolidDB as well. Started the SolidDB and then the VMGR.
Waited for the VMGR to finish starting, then I fired up the SP+CF server.

Bingo, Video Conferencing option was back!

I mentioned this to an IBM L2 support guy, and he will verify if this is actually so. He has a ST9 environment himself where he can test this.

Can anyone else verify?

There should be a "if no connection to vmgr, wait and retry" function on the SP+CF server. If there was, this would no be an issue.

Sametime 9 - HyperV - strange NIC issue, plus MCU NIC name must be a specific one

I had a strange issue today. A customer with a ST9 environment I previously installed, had no audio/video capabilities all of a sudden.

After debugging the Media Manager parts I noticed that the MCU and VMGR server had restarted for some reason (probably some admin doing maintenance), but I was not able to ping them.

Via the HyperV console, I got a hold of the console and, as suspected, no networking was enabled. And also, the number of NiC's was now 3. So I had "eth0", "eth1" and "eth3"...

And after I set the manual ip-address, I did a reboot, and then "eth4" was now present...
But when doing a "ifconfig" on the server, only the recently added nic was listed.

Why...?

Well, the reason was that the HyperV setting for the Ethernet card was set to "dynamic MAC address"!!!


If you're not a hacker, who would benefit from having dynamic Mac addresses, one might wonder. So to me, the option to set dynamic mac addresses is strange.

So, I found the MAC address for the "eth0" ethernet card inside of my RHEL 6.4 server, and then set the HyperV nic to this address. I also removed the newly created nic´s, so that only "eth0" was present. Restarted and all was good. I had the static IP back.

Side note:
The new network cards were automatically using dhcp, and no dhcp server was in reach, so that's why the network connectivity did not work. And also, using dhcp (if I had a dhcp server) would have give the servers the wrong ip address, since DNS record has been set manually.


Any-hoot, the MCU server needs the NiC-name to be "eth0". It won't work with any other name. I don´t remember where I read this, but it´s not inside of the ST9 documentation at all. I probably read it in the ST9 beta community somewhere...

So do remember to have "eth0" as the nic name on the MCU server!! You can verify this name by doing an ordinary "ifconfig" command in the terminal.