11-25-2013 03:17 PM - edited 03-16-2019 08:34 PM
Hi people,
I was wondering, (and I’m sure I DID read), on some CUCM documents that when we upgrade a firmware for a Phone, we have to upgrade the firmware in every CPE (Active Call Manager Node), and on the TFTP server, then do a wide cluster restarts.
Why does Cisco advise that if there is only a TFTP server and I backup TFTP server that some on the cases even don't run the Call Manager Service?
Sorry for this question,
I hope you guys borrow me some of your time, I'll appreciate it!
Solved! Go to Solution.
11-25-2013 07:45 PM
Manish is spot on. It is very key to understand that there is a difference between upgrading the firmware for an existing device type and adding a new device type. When upgrading the firmware of an existing device type, you only need to install the firmware cop file on the publisher node and the TFTP nodes. The former so things like Dev Defaults get updated. The latter because the TFTP nodes access their local file repository when serving files.
You have to restart the TFTP service so that the server can refresh cache. You DO NOT need to load firmware images on cluster nodes that are not running the TFTP service. In larger clusters, it is common to have nodes designated as TFTP only or as media only (or as TFTP + Media, if you walk that line).
When installing a dev pack, you typically need to update the db schema and update various web (admin/user) files (while accounting for new locales, etc.). Since the installer touches various file system locations, db tables, etc., Cisco opts to go for a UCM reload. Since you have to install dev packs on each server, that leads to the cluster reload scenario.
-Bill
Please remember to rate helpful responses and identify
11-25-2013 06:52 PM
Hi Juan,
One of the main reasons for installing the firmware on all the callmanager nodes is to take care of a situation where one of the primary TFTP server goes down and we need to use a backup server to serve tftp files. Also, installing a new firmware only requires TFTP service restart, a cluster reboot is required when we install a device pack on all nodes to add support for a new phone or device type.
HTH
Manish
11-25-2013 07:45 PM
Manish is spot on. It is very key to understand that there is a difference between upgrading the firmware for an existing device type and adding a new device type. When upgrading the firmware of an existing device type, you only need to install the firmware cop file on the publisher node and the TFTP nodes. The former so things like Dev Defaults get updated. The latter because the TFTP nodes access their local file repository when serving files.
You have to restart the TFTP service so that the server can refresh cache. You DO NOT need to load firmware images on cluster nodes that are not running the TFTP service. In larger clusters, it is common to have nodes designated as TFTP only or as media only (or as TFTP + Media, if you walk that line).
When installing a dev pack, you typically need to update the db schema and update various web (admin/user) files (while accounting for new locales, etc.). Since the installer touches various file system locations, db tables, etc., Cisco opts to go for a UCM reload. Since you have to install dev packs on each server, that leads to the cluster reload scenario.
-Bill
Please remember to rate helpful responses and identify
11-25-2013 07:45 PM
Hi Juan,
Agreed with Manish[+5].
You need to upload the desired firmware of IP phones in all servers in a cluster.In case of failure of any of the server where tftp service is running , the phones would switch over to another server in a cluster where CUCM service is running.In that case, this server should be updated with all running firmware on failed server.
You can also have a dedicated tftp server in a cluster.
regds,
aman
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide