Average Downtime when upgrading to 3.3

[Emily Mixon]

For anyone that has upgraded to Layer7 PAM version 3.3, how long did the upgrade take you to complete?  Along with that what version were you previously and roughly how big is/was your database?  Also did you encounter any specific issues when completing the upgrades or any information would be great.

I am trying to determine how much downtime to account for when upgrading our PAM instances. We have 3 VMs in our deployment that are currently running version 3.2.3.82.  Any help would be much appreciated.

[Andreas Mueller]

Hello Emily,

 

If your appliance is a VM and comes from r2.x chances are high you need to follow these steps to upgrade to 3.3

https://docops.ca.com/ca-privileged-access-manager/3-3/EN/upgrading/upgrade-to-release-3-3/upgrade-prerequisites-for-3-3#UpgradePrerequisitesfor3.3-UpgradeforVMwareOriginatingin2.x

which requires a few more reboots and time to complete the upgrade.

 

I suggest to plan one hour downtime per appliance in your cluster.

You can upgrade several appliances in parallel though.

 

Best Regards,

Andreas

 

[Joseph Fry]

Andrea's comment is perfect, however I wanted to note that there are some variables that may make it take longer than an hour.

• The amount of RAM and CPU resources your PAM appliance/VM has may impact the time to upgrade.
• The amount of data in your database has a large impact on the amount of time it takes to upgrade.

Recently I performed an upgrade on an appliance (former r2.x) with a relatively large database and only 8GB of RAM and just the upgrade itself ran for over an hour; total downtime was closer to 90 minutes.

[Julian Riano]

Hi Joseph


I have carried out the update process for 2 clients and the 2 have failed the first client due to the error of the following image

For the second client the error is as follows

Additional support CA requests to increase the RAM to 64Gb, this increase is disproportionate considering that it is a 400% increase in resources.

Why can this error occur and why is this large amount of RAM required ?

[Joseph Fry]

The amount of RAM would only slow your upgrade, not cause it to fail.  The 64GB RAM is a recommendation, not a requirement.  The recommendation usually reflects what our physical appliance ships with.  I have customers running 3.3 on 8GB of RAM.

I can only speculate what might cause errors like you are seeing.  I suspect you will need more detailed logs to actually determine what failed.  In the first case we know it was the schema update and in the second it was during a database restore.

Causes for these can be anything from corruption of data in the database to running out of disk space during the upgrade.  Open a ticket with support, they should be able to help identify the cause for the failed upgrades.