On Fri, 29 Sep 2006, Erik Kangas wrote:
Unexpected short mix message file        2441261 < 2441309

This is not a fatal error.

What has happened was that an expunge or checkpoint is trying to "burp" the data file (remove expunged messages). Before it does any writing, it makes certain that the file data ranges used in burping do not exceed the size of the file. If this sanity check fails, burping of that file is cancelled and neither the file nor the pointers into the file are altered.

In the example you gave, it exceeded it by 48 bytes. It is very strange that this happened, especially since 48 is much smaller than any plausible message size.

I wonder if there was some locking problem. What operating system are you using? Are you using NFS?

What does this mean?  Is it a fatal error, or is something else going on?

It's not a fatal error. Part of the design of mix is that the code detects that it's about to do something wrong and not do it.

Mix is also designed to be "self-healing". If a locking failure causes it to get inconsistant data between the index and the data files, it will not act on the inconsistant data; instead it backs off. Since this was just a burp, it doesn't really matter (other than failing to reduce disk space) that the burp did not happen.

If at a later time the data becomes consistant, then mix proceeds as if there never was a problem.

This doesn't mean that it's safe to do foolish things such as access mix over NFS (which disables locking). Mix dodges a lot of bullets; but it's best not to fire bullets at it in the first place.... ;-)

I went to the same folder today and tried to "mailutil copy" it to another folder and it worked without issues ... indicating to me that the folder seems "fine" at the moment.

Try making a copy of that mailbox with "cp -r" instead of "mailutil copy". Now, delete a message in the copy and do another expunge. Does the problem still happen? If so, I would be very interested in looking at the mailbox and analyzing what happened.

Thanks.

-- Mark --

http://staff.washington.edu/mrc
Science does not emerge from voting, party politics, or public debate.
Si vis pacem, para bellum.
_______________________________________________
Imap-uw mailing list
[email protected]
https://mailman1.u.washington.edu/mailman/listinfo/imap-uw

Reply via email to