During upgrades, our procedure is generally as follows:
- Perform undo checkout on any files that have been left checked out, this includes private state files
- If undo checkout fails (usually because the file has the same name as a vault file), delete from vault
- Backup PDM system
- Upgrade PDM system
Our procedure gives the vaults an annual cleanup and prevents files from being left checked out for many years. Of course there are many users who do not understand that local and private state files are not actually saved on the server. Depending on user-specific settings the undo-checkout/deleted files are either left local in user vault views or deleted automatically. We have never experienced great backlash from this procedure (from ~600 users), I suspect because most of these files have long been forgotten by the creating user.
I am curious whether anybody else performs this type of maintenance either during the upgrade or before/after. I cannot think of a technical reason why the files need to be checked in prior to an upgrade. Of course if all files are checked in, a subsequent server backup is ensured to contain 100% of data, including that which would otherwise languish only in user vault views. But the system will still upgrade, and the user data will still remain on their machine, if this is not done, correct?
Upgrade practices
Re: Upgrade practices
I leave private status files as is.
in fact they ARE added to the vault at creation, but could ve viewed by admin user and the user that made them only. (hence "private" or in the vault but not public)
I think the check them all in is a fail proof scenario if for some reason a client local cache is lost by accident during a failed upgrade or whatever...
I have 100 machines to admin and there is a lot of garbage with the same pattern you described.
GBs parked in some local cache unable to be checked in due to duplicate file name. and the person that should verify those left the company etc...
in fact they ARE added to the vault at creation, but could ve viewed by admin user and the user that made them only. (hence "private" or in the vault but not public)
I think the check them all in is a fail proof scenario if for some reason a client local cache is lost by accident during a failed upgrade or whatever...
I have 100 machines to admin and there is a lot of garbage with the same pattern you described.
GBs parked in some local cache unable to be checked in due to duplicate file name. and the person that should verify those left the company etc...
- Frederick_Law
- Posts: 1952
- Joined: Mon Mar 08, 2021 1:09 pm
- Location: Toronto
- x 1648
- x 1477
Re: Upgrade practices
I'm using Vault, not PDM.
I setup a "Checked Out By" search for users (only 2 LOL).
We're supposed to check at least weekly.
And before vacation.
I setup a "Checked Out By" search for users (only 2 LOL).
We're supposed to check at least weekly.
And before vacation.
Re: Upgrade practices
Thanks for the replies. We performed our usual procedure 2 weeks ago and there was no screaming. Nice clean vaults!