Bill Long writes:

Quick question

I'm getting continuous message like
Aug 15 17:58:08 ns courierd: QUEUE FILE CORRUPTION: inode 535350
Aug 15 17:58:08 ns courierd: QUEUE FILE CORRUPTION: inode 535341
Aug 15 17:58:08 ns courierd: QUEUE FILE CORRUPTION: inode 535352
Aug 15 17:58:08 ns courierd: QUEUE FILE CORRUPTION: inode 535355
Aug 15 17:58:08 ns courierd: QUEUE FILE CORRUPTION: inode 664408

in my logs. I have identified the files with the corresponding inodes. they are all in the tmp directory.

Can i just blow them away? or will that mess all kinds of stuff up? I saw a few past posts where Sam recommended to stop the system, blow the entire queue away and restart it. I'm hoping that was just to shotgun the bad files.

No. The reason they're in the tmp directory is because some other message has the same filename in the msgs or the msgq directory, hence when Courier tries to move the new message from tmp to msgs and msgq, it finds that some other file is already there.

Since the files are named according to the inode number, either the files in tmp have a wrong name, or the files in msgq and msgs have the wrong name.

stat the files in tmp, and remove them only if their name is different than their inode number, as shown. But chances are that it's the msgq and msgs files that got corrupted. Fix them as follows:

You need to:

1) courier stop

2) find msgq -name [CD]535350* -print

3) find msgs -name [CD]535350* -print

4) Blow away all files found by the above.

5) Repeat for the other inodes

6) courier start



Attachment: pgp00000.pgp
Description: PGP signature

Reply via email to