In <20010608094124.A31605@???> ivan.beveridge@??? (Ivan Beveridge) writes:
>(I was going to reply to Don's reply to this, but I couldn't be bothered
>with shuffling around the mail and adding the missing quoting :)
>On Thu, Jun 07, 2001 at 02:06:19PM +0100, John Burnham wrote:
>> > Are you, by any chance, running qpopper? I have seen this happen once or
>> > twice on such (thanks to the Lines: command being mangled, iirc - I dont
>> run
>> > qpopper)
>> >
>> Yeah, I am actually. Hmm, on my previous mail server I didn't have any
>> problems with it. Oh well, I have changed my POP3 server and will see if the
>> problem goes away. Thanks to all those who helped.
>Don mentioned that he had the problem, but used uw-imap, and moving to ext2
>fixed it. If there is still a problem with reiserfs, this is a bit of a pain,
>as I would want to use some kind of journalling FS (I guess xfs or possibly
>jfs is an option).
>Anyway, I was wondering if using maildir as a format, rather than normal
>mbox, would alleviate the problem. Obviously it doesn't fix the root of the
>problem, but just tries something else. I'm not sure how you could easily
>convert from mbox to maildir, but it may be an option.
The description of the problem looks to me like a file locking issue. I
have seen the same thing when a home brewed pop daemon at a company I
worked for had got file locking wrong. I think it would be very safe to
say that using Maildir format mailboxes would definately fix the
problem, as, as you not, Maildir keeps each message in a seperate file,
so the chance of intermingling messages is eliminated.
I haven't had much time to look at Reiserfs. Is there any test data on
how efficiently it handles lots of small files in a directory as
compared to ext2fs? Ext2fs is fairly poor at huge number of files in a
directory. This was likely to be more of a problem with news than mail
though. Another nice feature you get with Maildir is a simple to
implement ability to expire read mail for your users, if your policies
allow you to do this.
--
Artificial Intelligence stands no chance against Natural Stupidity.
GAT d- -p+(--) c++++ l++ u++ t- m--- W--- !v
b+++ e* s-/+ n-(?) h++ f+g+ w+++ y*
