PDM, corrupt Archive folders in archive server.

[bnemec]

Has anyone dealt with corrupt archive folders in the archive server? I am seeing this only on the hex-decimal of doc id named folders (contains thumbs and versions of ONE vault file). The 0 - F folders are all fine.

Trying to browse into the folder on archive server gives typical windows corrupt directory error.

I'm at a bit of a loss and searches have all just lead to physical media errors. The archive server is a VM so disc corruption could be possible, but would be identified by the hypervisor. From what I can understand the failed media would be detected by some layer between physical media and the "drive" on the VM. That is out of my domain as IT manages the storage and VM farm. I just have two servers out of dozens on the stack.

[jcapriotti]

I have a few years ago but don't remember if we were able to do anything about it. Maybe some sort of disk scan and repair like "chkdsk"?

[bnemec]

I have a few years ago but don't remember if we were able to do anything about it. Maybe some sort of disk scan and repair like "chkdsk"?

I started with Windows drive scan tool which said it couldn't finish the online scan and to try offline scan. Which I can try tonight.

Chkdsk was executed in scan mode on a volume snapshot.

Checking file system on D:
Volume label is New Volume.

Stage 1: Examining basic file system structure ...

2455552 file records processed. File verification completed.

606 large file records processed. Found unused file metadata marked as used
... queued for offline repair.

0 bad file records processed.
Stage 2: Examining file name linkage ...
Read failure with status 0xc000000e at offset 0xc1360000 for 0x400 bytes.

Snapshot was deleted. CHKDSK cannot continue.
A snapshot error occured while scanning this drive. You can try again, but if this problem persists, run an offline scan and fix.

[mp3-250]

is the archive on a RAID array?
have the windows log files recorded a disk surface failure or other disk related errors?

[AlexB]

Have you checked the System and Event Logs for the Archive Server? They may shed some light on what may have happened. I haven't dealt with anything like that myself but it may benefit you to restore just that folder from a backup drive, although you'll potentially lose some progress.

[bnemec]

Have you checked the System and Event Logs for the Archive Server? They may shed some light on what may have happened. I haven't dealt with anything like that myself but it may benefit you to restore just that folder from a backup drive, although you'll potentially lose some progress.

I had not, but that's a good idea. I don't know exactly when the folders became corrupt.

As for restoring archive folder, a little loss of work is not as concerning to me as creating an inconsistency between SQL and archive for the document.

[AlexB]

As for restoring archive folder, a little loss of work is not as concerning to me as creating an inconsistency between SQL and archive for the document.

I've worked with a test database that I restored from a backup of the production database. In doing so, I ran into the issue of the SQL and the Archive folders not matching. I was able to delete/rollback the problem files or the files I didn't want in order to remedy this. You could also pull the matching version from the backup and overwrite the problem files, but I imagine this process to be tedious unless you create a RoboCopy script that could do the copying for you. I'm not saying it's an easy fix, but it's possible.

That's frustrating and I definitely feel for you.

[JSculley]

I had not, but that's a good idea. I don't know exactly when the folders became corrupt.

As for restoring archive folder, a little loss of work is not as concerning to me as creating an inconsistency between SQL and archive for the document.

If you haven't seen it, there is a detailed article in the KB on how to manually recreate the index.xml. It can be helpful when dealing with recovering from corrupt files/folders.

[bnemec]

If you haven't seen it, there is a detailed article in the KB on how to manually recreate the index.xml. It can be helpful when dealing with recovering from corrupt files/folders.

That's perfect, thank you. I started reverse engineering the xml files in the archive of other files but this detailed doc is much better.

[bnemec]

I was surprised to find that we can manually "cache" files on client workstations by working offline, then copying the file into the correct folder, then switching back to online. Most of the files that have corrupt archive folder on the server archives are cached in various user workstations so we have copies of the files.

As for restoring the archives it looks like our least bad option is to try use force switch on DEL command on corrupt folders and then manually recreate them using files in user cache and the info in the pdf in the KB that JSculley referenced. We won't have old versions of those files, but will at least get normal caching back.

[AlexB]

I was surprised to find that we can manually "cache" files on client workstations by working offline, then copying the file into the correct folder, then switching back to online.

To simulate this without going offline then online a bunch, you can access the cache by simply adding two backslashes to the path in the navigation bar. I forget where I saw it but I'll definitely remember it.

See GIF

Edit: I tested it and it only allows you to delete the cached items. It would not let me add any back in. Perhaps going offline is best then.

[bnemec]

To simulate this without going offline then online a bunch, you can access the cache by simply adding two backslashes to the path in the navigation bar. I forget where I saw it but I'll definitely remember it.

See GIF
Cache access.gif

Edit: I tested it and it only allows you to delete the cached items. It would not let me add any back in. Perhaps going offline is best then.

That is a neat trick! Sometimes I just want to see the regular file properties of a vault file instead of the data card. This will be handy for that. Thanks for sharing.

[mp3-250]

@bnemec you can also access the filesystem of the local cache using an utility like totalcommander, double commander or the command prompt.
I made a batch file to convert local cache to local folders and viceversa, renaming the desktop.ini in the root folder of the vault.