Odd workspace behavior after "clear cache"

[egifford]

We’re running Wildfire 4 M230 with Windchill PDMLink 10.0 M030. We’ve experienced some odd issues after using the “clear cache” function under Tools, Server Manager, Cache Tools. After running this, the user activates a workspace, synchronizes and applies a check out to a file – workspace display updates to include the “checked out” symbol in that row. Open that file and attempt to modify it. Pro/E throws a conflict to the screen indicating the file is read only as it is not checked out and offers the options of check out, continue, revise and check out or cancel. If the user selects check out, it errors out saying the checkout fails because the file is already checked out. ??? Event manager shows “the object has already been checked out by an unknown user”. If I return to viewing the workspace, indeed the file still shows the indicator for “checked out” to that workspace. Attempting to cancel the check out fails with “undo checkout failed. See the event console for more details.” This still leaves the file displaying as checked out. If I synchronize the workspace again everything now starts to behave. Why do I have to synchronize after applying a check out after using clear cache? Similar behavior occurs if the workspace had files that were already checked out but were unmodified before the clear cache. Why does it behave this way?

I opened a call with PTC and the tech’s answer on the phone was to NOT use the clear cache function. PTC recommends checking in and / or uploading all new or modified files you want to keep, deleting any workspaces you can, exiting Pro/E and renaming or deleting the .wf folder. This then forces Pro/E & Windchill to start with a totally clean cache folder structure. He seemed to indicate the use of the clear or clear cache functions from Server Manager is not really “supported” by PTC. Of course he did not document any of those statements in the case number. It looked initially like the clear cache functions would alleviate some weird lockup issues we were having when working in Pro/E and accessing Windchill PDMLink through the integrated browser panel. However, resulting issues with basic functions like check out make me have second thoughts about the clear function.

Has anyone seen odd behavior like this? Have any recommended best practices? I strongly advise my users to keep workspaces to a minimum, use them for a short time for specific task etc., then delete them. However the cache can still grow fairly large which I believe leads to the flakey issues / hangs in interacting with Windchill within Pro/E. Hope was clear would be the magic bullet.

Thanks!

Erik

[dschenken]

It seems to me like the cache is where the Commonspace status of items is held locally, so clearing it removes those, like a lobotomy. The way to restore it is to synchronize. I'd guess cache clearing is of benefit when synch doesn't work for some reason.

Software typically fails because it is poorly written, not because there's too much to do. If it is poorly written you can cope by keeping the loading on it small, but imagine suggesting that no one re-use a spreadsheet, just start from scratch, because if they get used too much they get corrupted.

The best practice I can think of is to minimize the information that the Workspace seeks from Commonspace by limiting columns. Each column requires database accesses for each component displayed. I keep mine to a minimum by only displaying items that I have checked out, unless there is some reason to get more info. Instead of updating info for 10,000 items in a workspace, it only needs to get info for 2-3.

There has never been a reasonable explanation of Workspace corruption issues. My guess has been that some processes time out and leave records incomplete, which is why asking for fewer updates has done well enough. I've had workspaces between 20 and 50 times larger than the maximum recommended size working correctly for months at a time - if it was a size problem, I should have seen failure daily.

What baffles me is why PTC puts all the users eggs in one fragile basket. Workspace should be entirely independent file/disk structures. I should be able to use Windows Explorer to delete a local Workspace and, except for no longer finding it, be able to work with any remaining workspace.

I'll again mention the PTC development cycle should start with a description of how things are to interact, not a circus parade follow-up sweep of what hit the street and where.

[egifford]

Thing is we syncrhonized the active workspace right after clearing the cache and before starting anything in Pro/E. There are a number of posts on the board about using the clear cache as a recommended practice & it's in PTC's help docs, so I was a little surprised to have tech support basically tell me "yeah, we really don't recommend you use that function. You're better off nuking the user's local cache folder and starting fresh.".

If I'm using the clear function incorrectly, I'd like to know how to use it the right way. From posts and the help pages it looked like it would be a nice tool users could run on a regular basis to keep their PDMLink session, well, "regular".

[dschenken]

There are two Workspace versions - one on the server and the local one. I don't think you can do anything to the local one except through Pro/E, so if you clear the cache you have to synchronize via Pro/E.

I'd refer to the documentation, but there doesn't seem to be any concerning detailed operations**. What I do know is that this is the way it appears to work.

**It's as if the owners manual for a car only described how to use Google Maps, because there's no chance the car will break down.

[egifford]

This is getting more strange. I received another reply from tech support regarding the Server Manager tools for clearing the client side Windchill cache. To me the reply seems to not only contradict the functionality as it has been discussed in threads on this board, it contradicts PTC's own help pages, documents from the knowledge base and what I've observed when I run the tools.

The latest from support is:

"The Clear cache functionality in the cache tab of the server management window does not clear the local client cache. It only helps to clear the login details and basic details from the cache.

PTC recommends customer to rename the cache directory (.wf ) if they wish to clear their local client cache."

Unless the functionality has changed with newer releases / software combinations, and that is what support is refering to, there seems to be some conflicting information regarding what the cache managment tools and settings actually do. I was going to recommend that users use the cache tools off of the Server Manager interface on a periodic basis, but it seems PTC recommends starting with a new cache directory instead.

Can any other admins out there chime in with their experience? How about anyone from PTC regarding "official" recommendations?

Here are some screen captures from PTC's help pages in Pro/E and from the knowledge base. There is more in Windchill documentation, too.

[TomU]

I have used the cache cleanup tools extensively during performance testing and never had any corruption issues at all. I also have never needed to synchronize after running a cleanup. In my experience the cache will "auto clear" for simple, recent operations.

For example, if you create a new workspace, add several thousand objects, and then either delete the workspace, or delete all the objects in the workspace plus all the frames (events) related to that workspace, the recently added objects will automatically be cleared out of the cache. You can see this by watching the cache size after each operation.

The cleanup tools seem to be more necessary for things that have been left there for a long time or accessed by multiple workspaces. Keep in mind that the cleanup tools will NOT remove any object still listed in any of the workspaces or referenced by any frame (previous event). This means you are really just removing earlier iterations or versions no longer accessible by an earlier frame.

I am not a big fan of creating new caches primarily because I have many different workspaces containing many different objects that are either new and temporary (sketches, udfs, etc.) or contain modified, non-checked out objects that are not eligible for upload. Creating a new cache means I have to go through all those workspaces and do something with all these temporary objects, and checking in may not be an option.

I would say the vast majority of our users have NEVER recreated their cache in almost 2 years of operation. I've had my cache grow to well over 16 GB with no issue whatsoever.

(WC9.1 M050 + WF5 M120)