I need to upgrade from 5.1 to 5.1(3b) due to an issue with disk space on the active partition - I am down to 2% free space. This is a bug I guess - seem to get different answers from TAC.
Anyway my question is this. I have multiple sites running off of a single cluster at my HQ. The connections to remote sites is MPLS (1.5mbs). There are 3 remote sites - about 300 phones remote. WHen I do the upgrade I upgrade the Publisher first, then the Subscriber. So when the Subscriber upgrades - I assume the phones register with the Publisher and will that cause all the phones to do a phone load upgrade. My conscern is what happens if multiple phones are trying to get to the TFTP server for new downloads at the same time over the WAN. Is this an issue??
All the phones loading together could overload the WAN links. Try disabling all the phones and bring them up in groups leving time for each group to load before enabling the next.
mconnely in 6 you can choose to not reboot at the begining, and leave the machines in the active partition. do this, upgrade them all, publisher first, then subscribers.. and then reboot them all at once into the new partition with new CM versions... not sure if this is an option on 5? This way your phones all get the new load from the subscribers instead of pub. Also, are all your subs TFTP's as well?
Chad, I thought Bill was more concerned with flooding the WAN and possible timeouts on multiple tftp loads. Anyway that is a handy option to leave the phone booting till later, I don't know if it's in 5.
just as a side note, you do not upgrade pub and then sub. you install upgrade to ALL servers and then reboot them in order switching the partitions
if this helps, please rate
Thanks - that is what I did - install on each and then switch. As it turns out the downloads across the WAN took about an hour on each link - approx 100 phones on each. My MPLS did a nice job with it - I could talk across the maxed out link with very little drop-out.
It's the same in 5.x - you upload the update file and then tell it to install on the inactive partition but not reboot automatically. Then you can go to each in turn and switch versions. In my case I have a publisher and 1 sub located at my HQ site. The remote site's phones are all registered to the sub. So when you reboot the pub, the phones don't see that. WHen the sub switched then all the phones did their downloads. I'm not sure if phones tried to register with the pub while the sub was re-starting or if they were still trying the sub and then once it was back up they registered with it and did their downloads. When everything was done - say an hour after the switch, all phones were registered to subscriber. Out of 580 phones we only had to manually re-boot 5.
I did an upgrade from 18.104.22.1680-11 to 22.214.171.12400-4 over the weekend and for most phones the call manager lost some detail in the Directory Number Configuration. What went missing were the data in the "ASCII Display (Internal Caller ID)" & "ASCII Line Text Label fields". Before I go through the rigours of raising a TAC case I'm just wondering if you had a similar experience? Most of the phones are the 7912 type so due to the lower spec display type they use the ASCII data fields.
Another strange thing was a backup only shows the status of progress when it is finished.
I'm on vacation so I might need to check further, but no reports from users about 7912's. Backups show success for CCM. I did notice that it doesn't show success for CDR since the upgrade - will have to check on that.
Anymore info about this. Had a customer report this problem to me today. 7920 and 7985 phones not displaying the caller name, just ext. 7971 phones work just fine. Looked and noticed that all the ASCII fields were gone.
The only update I can give you is that we put back all the ascii text fields but didn't pursue it with TAC. It definitely looks like a bug ok.