I'll be migrating a Unix Perforce installation to a new (Unix) server with more storage next month. Because we now have a lot of Windows users, I'd like to start up the new server in case-insensitive mode (-C1). According to Perforce, this scenario isn't supported (http://answers.perforce.com/articles/KB_Article/Cross-Platform-Perforce-Server-Migration).
My question: Assuming I can verify that there are no case conflicts among current files and directories in the depot, would restoring a case-sensitive checkpoint using case-insensitive mode work? What if there are deleted files and directories that contain case conflicts? what would happen if a rollback was attempted to a time when case conflicts existed?
As I see it, my only other option for keeping the file histories accessible would be to start the new server with the current file versions. Then on the existing server I'd delete most of the existing user accounts, transfer the license to the new server, and keep the old server around (indefinitely) under the free license model.
Thanks for any help!
Finally got around to preparing for this migration, and I can now state definitively:
p4d will NOT allow you to restore a checkpoint with case-insensitive mode turned on (-C1) if that checkpoint was created in case-sensitive mode on a Linux server.
When you try, it will report an error that the checkpoint was created in case-sensitive mode.
Related
I hope I'm not asking a question that's already been answered, but I can't seem to find one that fits my situation.
Scenario: Using P4V gui (2011 version), with no access to P4 command line, on Windows 7.
The setup: A user creates a workspace in Perforce from Machine A, pointing that workspace to a shared network drive, and checks out a file for editing.
Machine A then dies before the user can check in the file. The user is then assigned Machine B, for which he must create a new workspace (which is also pointed at the same shared network drive).
The problem: The problem we're having is that even though the workspace from the dead Machine A and from the new Machine B both point to the same location, Perforce considers them to be different workspaces and prevents the user from checking in/submitting the previously checked out file.
Any suggestions on how to check in this stranded file would be greatly appreciated. Thank you very much!
To be nice and clean, I'd suggest this:
Make a backup copy of the file as it exists on the shared network drive.
Connect to the depot from Workspace A (can be done from Machine B if the workspace isn't bound to Machine A).
Revert the file - this will overwrite the file on the drive with the version in the depot (aren't you glad you made a copy first?).
Switch to Workspace B.
Check out the file.
Copy the backup version over the file on the shared network drive.
Check in the file.
...and if you're not planning on using Workspace A anymore, I'd suggest deleting it.
Have you tried clearing out the Host field? For example, see 'Using the same workspace from different machines':
http://www.perforce.com/perforce/doc.current/manuals/p4guide/chapter.configuration.html#d0e1720
As an update from my comment response, if you need to change ownership of a changelist, the steps are documented here:
http://answers.perforce.com/articles/KB_Article/Changing-the-Owner-of-a-Pending-Changelist
Recently our development team received new pc's. In an effort to make this transition smoother, I would like to be able to explain to my co-workers how to continue using the client they already have set up to pull files to and from the new pc while eventually ignoring the old pc workspace altogether.
I know about adjusting the attributes of the client itself and allowing the client to be accessed by different hosts. What I'm looking to do now is update the perforce have list for the given client to reflect the files (or lack thereof) that are on the new pc's file system (in the correctly mapped location, obviously).
I'm not sure if it is possible with the p4 flush command for perforce to know which revision of an existing workspace file i have without explicitly telling perforce which revision it is...? (this seems like its asking a lot)
Apart from files that Do exist in the workspace, is there a command that will update the have list to #0 for files that don't exist in the workspace?
OR
Is the sledgehammer approach:
submit any pending changes in the old and/or new workspace
remove any files that may have already been forced into the (new) workspace
$:p4 flush [workspace root]/...#0
appropriate in this situation?
If using the existing workspaces is an option, then this should be pretty easy. It sounds like you already know how to make a workspace accessible from a different host (you can leave it blank to make it accessible by any host). If you copy the workspace folder to the new PC, and update the root of the workspace as necessary, it should "just work" without any additional changes.
If I'm understanding your question correctly, I believe that using a workspace name as your revision modifier will do what you want. For example p4 flush //depot/path/some/file#workspacename. For new machines, we often go through these basic steps to avoid having to resync files.
Copy the files in the workspace from machine 1 to machine 2
Create a client that matches the old client's mappings
In the new client, run:
p4 flush //depot/...#oldclientname
I forgot to check out a source code file before modifying it.
When I get last revision, Perforce overwrote that file, so my work is totally lost.
Is it possible to recover the file?
For future use, update your client workspace so that you specify "noallwrite, noclobber". If noclobber is set, Perforce will not overwrite your writable un-opened files: http://www.perforce.com/perforce/doc.current/manuals/cmdref/client.html
Only if your editor or your operating system saved a copy or it's been modified long enough that it made its way to your backups. Perforce will not make copies of such files, it blindly assumes that you didn't lie and will always honestly tell it when you want to edit a file.
if you are using eclipse then its possible to retrieve the local version using Compare With -> Local history . It helped me.
This has happened to me recently. For some reason, after I "p4 sync"-ed my workspace, and do p4 resolve, I noticed that my changes to a file were missing. I'm not sure if my changes were not saved or I haven't checked out the file. But I really remembered that my changes were saved. :(
I have been using Visual Studio for development and it doesn't have local history unlike in Eclipse. Luckily, that file is a javascript file and I have been testing my application in Internet Explorer. Since IE does some caching on some internet data like js files, what I did is to check the directory where it saves temporarily files (Internet Options -> Browser history settings ) there you'll see different versions of the files saved. I did recover my files! It was really just luck!
After that incident, I installed a plugin for visual studio for storing local history of files everytime it's being saved. http://visualstudiogallery.msdn.microsoft.com/226c2108-9da9-407d-b90d-9783040d27b8
Best thing to avoid these cases is to:
branch out your files first into a separate devline during development and submit every milestones you accomplished
incrementally. In this way you'll always have versions of important
changes you do during development. After this you could
integrate it back to the parent branch/mainline.
http://answers.perforce.com/articles/KB_Article/Branching-Codelines-and-Merging-Changes
Hope this helps!
If you fired the following command (which is a FORCE sync option), only then will Perforce update ALL your files.. including ones which are WRITABLE. The only exception is that any file that you have OPENED in perforce will not be overwritten. So if your file was made WRITABLE using OS command, and not using p4 open.. they will get overwritten by p4 sync -f.
p4 sync -f
The other possibility is that you did p4 sync, and still perforce overwrote your writable files (which were not opened using p4) because your workspace settings don't have noallwrite, noclobber specified. Usually by default, these settings are already specified, so that Perforce doesn't clobber writable files.
I tryed migrate from perforce 2005 to 2010 and failed. I created checkpoint and journal files (with p4d -jc command), copy version files on old machine. But failed restore it on new machine. On first try i get
"Journal file 'journal' replay failed at line 9680!
Bad opcode '' journal record!"
error. I found in journal file zero bytes and remove it. On second try I get error:
"Perforce server error
1 out of sequence journals were not replayed"
What do next? How to repair journal file? Can I restore Perforce database without journal file using only version files?
P.S. Perforce still work on old machine.
Your journal got corrupted at some point. Since perforce is still running on the old machine I suggest you to start the migration over ensuring at each step that nothing got corrupted. For instance look at the journal size after generation, or better make a md5 or an sha1 of it, and check everything is still ok on the new machine. Also ensure that p4d is stopped before copying any file.
Since you upgrade from p4 2005 to p4 2010 you probably have a specific procedure, did you follow it?
My home Perforce server died. I set up a new one.
The project I set it up to support died in the planning phase. The contents of the depot at that point were some prototype code and we never got to setting up a disaster recovery plan.
The dev machines still have the existing code on them. As much as possible, I'd like the change of servers to be transparent to the developers--use the same depositories and the same directories, just change the name of the server to connect to and get back to work.
What do I need to do in order to make this happen?
I assume you don't have access to the perforce depot files from your dead server? I assume you know that you will lose all your history.
If that's the case all you need to do is setup the new server, create a user / client with the same root clientspec path as your original clientspec was using on your dev machine and checkin all the files into perforce. Pretty simple really...
You may need to rebind is SCM binding that you may have in tools like Visual Studio but that's about it.
What Shane suggested will populate the depot with one person's version of the files. But if you have another user who also has a copy then you'll need a couple of extra steps.
Firstly, just set one machine up as suggested by Shane.
You now need to get the second user set up. If you are confident that the version of the code user 2 has exactly matches what you put in the new server, then just create a client spec (probably same name as used before), and then sync using the "Force" flag. This will overwrite all the files on user 2's machine, and - more importantly - ensure Perforce knows which versions you really have.
However, if you are in any doubt as to any differences in code, then do not do the initial sync from the second user's machine. Instead, set up the client spec, then use the "Reconcile offline work" option - from P4V select the workspace, then it's a right click option. Then just walk through the subsequent dialog to sort out what you need.
Finally, if you want a very quick & dirty backup system for your server, I've posted some notes on my blog here - should take you just a couple of minutes to set up.