I have a problem during upgrade process for unityconnection cluster from version 8.5.1 to 8.6.2 .
I installed on both of the servers (Pub&Sub) the ciscocm.refresh_upgrade_v1.1.cop.sgn and then I insert the Installation CD to the CDrom of the server .
after two hours of installation the system goes down for restart and this is the output prom putty :
Successfully installed UCSInstall_UCOS_188.8.131.5200-2.sgn.iso
Refresh Upgrade prepare processing was successful. The system will reboot shortly to continue the Refresh Upgrade. Refresh Upgrade progress can be monitored from the system console. Once CLI and GUI access is restored, upgrade completion can be verified.
Then the system start to install 8.6.2 version and after couple of restart when everything look ok the system shows the next prompt (on the blue screen) :
"error detected in post initial phase. trying to restore the system to the active partition"
Then the system come back to 8.5.1 version with no inactive partition from "show cuc version" :
admin:show cuc version
Active version: 8.5.1ES65.13900-65
I try to upgrade three times and every time I get the same prompt .
Please , someone can assist me ?
Thanks alot ,
After you upgrade the Primary UC inactive partition make sure you do not attempt to switch partitions until you have upgraded the inactive partition on the Secondary. What I suspect is if you reviewed the upgrade installation logs that both inactive partitions version don't match when the attempt to switch versions is made on the Primary.
So to recap, upgrade the Primary inactive without switchover, then upgrade the secondary inactive partition without switchover. Once the secondary upgrade is completed then initiate the switchover. My recommendation is to use the CLI to initiate the switchover, that you can monitor it more closely.
Please rate helpful posts. Thank you
Sent from Cisco Technical Support iPad App
Thanks for your answer . I try to made what you told me to do but after uploading the file I dont have the choise between the new version or the old version . please see the picture below :
thanks alot ,
Thanks for the clarification, the reason that there is no switch-over command is that the this a refresh upgrade as mentioned in your original post, what I mentioned in my previous post still applies if you we're upgrading between SUs in a UC cluster. However, in this instance there are certain requirements that need to be adhered for a refresh upgrade, the main specific requirement is that the hardware requirements are met. Depending on the hardware there maybe a number of upgrades, which will result in the active partition being overwritten.
In your case the inactive partition is empty due to the failed upgrade, if the upgrade fails this is part of the cleanup process. We need to determine why the refresh upgrade fails and could be due to a number of caveats that need to adhered to for upgrading from 8.5. Depending on the MCS or HP equivalent hardware the upgrade converts the hardware raid to software, please refer to the below:
About Upgrades to Connection 8.6
Note the following considerations about upgrading a Connection server to version 8.6:
•If you have never upgraded the Connection server before, the upgrade copies the new version to an empty partition.
•If you have upgraded the Connection server before, the upgrade copies the new version to the inactive partition, which usually contains a version of Connection older than the version that is running on the active partition. (If you previously upgraded to a later version and then reverted to the older version, the inactive partition includes a version of Connection later than the version that is currently running.) The software that was on the inactive partition before you started the upgrade is overwritten.
•Depending on your current version, you may be required to upgrade twice to reach the desired version. In that case, the current version is no longer available when the upgrade is complete because the partition that contains the current version will be overwritten by the second upgrade.
•If you are upgrading a Cisco MCS 7825-H3 server (or the equivalent HP DL320G5), Connection must reformat the hard disk to convert from hardware-based RAID to software-based RAID. When the upgrade is complete, only the latest version of Connection will be available on the hard disk.
•While upgrading an MCS 7825-H3 or HP DL320G5 server, Connection saves data and voice messages to an external drive. You must have a 128 GB or larger USB flash drive or external hard disk to perform the upgrade for these servers.
Can you confirm what your hardware platform is that you are attempting the upgrade on, I have provided a link to a document which outlines the above, and other considerations for upgrading from 8.5:
Below is also a link to the support hardware requirements, platform, memory and hard disks for your reference:
If possible can you retrieve the upgrade logs from the inactive partition through the RTMT as this should provide us with an indication as to why the refresh failed. Incidentally do you have NIC failover enable? There is a known issue with refresh upgrades from 8.5.1 on 7825H3 or 7828H3 failing if failover is enabled, for you reference:
CSCts35174 Bug Details
RU 8.5(1) to 8.6(1) fails on 782xH3 when N/W Fault Tolerance enabled.
Upgrade from 8.5.1 to 8.6.1 fails.
1) Fresh install of 8.5.1 on a 7825H3 or 7828H3
2) Network fault tolerance is configured on server
3) Upgrade is attempted from 8.5.1 to 8.6.1
Disable network fault tolereance prior to upgrade. After upgrade is completed, Network fault
tolereance can be configured again on server
Sent from Cisco Technical Support iPad App