February 2010 - dovecot

[Dovecot] LMTP in 2.0: concurrency, inet sockets
by Stefan Foerster 07 Feb '10

07 Feb '10

I didn't really have much time to look at the beta versions of dovecot 2.0 (apart from installing it once and migrating most of my current configuration, purely for testing purposes). An exciting new feature is the LMTP support. In the default configuration I have seen, that one only listened on a Unix socket - will it support network (inet) sockets, too? What about concurrent deliviers? As of now, when using the deliver LDA together with Postfix, you have to make sure that only one recipient is passed to deliver at a time, so for multi-recipient mail, the system might spawn a large number of processes (two per recipient, therefore two forks/execve calls). Is that limit still imposed with the new LMTP server? Stefan

3 4

Re: [Dovecot] LMTP in 2.0: concurrency, inet sockets
by Timo Sirainen 07 Feb '10

07 Feb '10

On Sun, 2010-02-07 at 13:40 +0100, Stefan Foerster wrote: > For 100k messages, delivery to a Maildir store using Postifx's pipe(8) > and Dovecot's deliver took 4319s, which equals 23.15 delivieries per > second - and that's with the pipe(8) delivery agent haning around for > up to 100 seconds, being reused for multiple transactions. > > Dovecot 2.0, with LMTP, was finished after 4007s, which equals 24.96 > deliveries per second. You could also try if maildir_very_dirty_syncs=yes affects it.

1 0

[Dovecot] blocked for xx secs
by Andrew Hearn 06 Feb '10

06 Feb '10

Hello, Occasionally (a few times a day) we get batches of Warnings in the log such as: dovecot: Warning: chdir(/maildir/x.co.uk/users/y) blocked for 18 secs This tends to last a minute or so for different users. /maildir is a NFS mount, and I'm suspecting it's the NFS server being busy and dovecot having to wait. Other than replacing the NFS server with faster hardware is there anything else we can do? Indexes are local, and Maildir is on the NFS. Dovecot 1.2.5 using: mail_nfs_storage = yes mmap_disable = no NFS mounted as; /maildir on /.automount/a.diskless/root/mail type nfs (nosuid,nodev,noatime,vers=3,proto=tcp) TIA -- Andrew

2 1

[Dovecot] Compressing Maildir mails on delivery
by Timo Sirainen 06 Feb '10

06 Feb '10

Now supported by v2.0. Also as a patch to v1.2: http://dovecot.org/patches/1.2/zlib-compress.diff I'm not really sure if I should commit it to v1.2 code tree. The code contains ugly copy&pasted io_stream_copy() and v1.2 is supposed to be feature complete..

3 5

[Dovecot] Also - is there a reliable way for TB to know it is connecting to a dovecot server?
by Charles Marcus 06 Feb '10

06 Feb '10

This would be so it could auto-disable GLODA - its new 'Global Search and Indexing' scheme - for that account - assuming they accept (and implement) my bug request to be able to do that: https://bugzilla.mozilla.org/show_bug.cgi?id=544100

2 6

[Dovecot] upgrade 1.0.x -> 1.2.x
by maximatt 06 Feb '10

06 Feb '10

its posible upgrade from dovecot 1.0.x to 1.2.x? i only see upgrades over micro versions.. thanks! -- Salu2 ;)

2 1

[Dovecot] v2.0: tcp-wrappers support
by Timo Sirainen 06 Feb '10

06 Feb '10

Some people have been requesting this forever. So finally this works: service tcpwrap { unix_listener login/tcpwrap { mode = 0666 } } login_access_sockets = tcpwrap There can also be more sockets listed, in case someone wants to build their own access checking service.

1 0

[Dovecot] pop3 quick question
by Nikita Koshikov 06 Feb '10

06 Feb '10

Hello list, I have found imap_allowed option in dovecot which do exactly what I need, for imap. Is there something similar for pop3 protocol? The main goal for it - is to enable pop3 access to specific users. Or maybe dovecot have some 'pop3=yes' or 'imap=no' args to userdb/passdb sections ? I know future dovecot 2.* versions will support rewriting for protocols, but this is required for 1.2 branch.

2 2

Re: [Dovecot] mdbox compression
by Damon Atkins 06 Feb '10

06 Feb '10

ZFS has support for compression on the file system ( lzjb | gzip | gzip-N | zle ). gzip eats CPU even at levels as low as 3. From the zfs man page The lzjb compression algorithm is optimized for performance while providing decent data compression. You can specify the gzip level by using the value gzip-N where N is an integer from 1 (fastest) to 9 (best compression ratio). The cost of CPU to support compression (gzip) vs the Cost of Disk if mboxes are not compressed? The other interesting feature ZFS (OpenSolaris) has is de-duplication, it detects if the block which is about to be written, is the same as another block and instead of writing the block it puts a pointer to the already existing block, saving disk space. Only problem with this is you would need to filter out headers which are different even when the to/body is the same. Cheers

1 0

[Dovecot] mdbox compression
by Timo Sirainen 06 Feb '10

06 Feb '10

I was wondering if I should add compression support to mdbox one mail at a time or one file (~2MB) at a time. The tradeoffs are: * one mail at a time allows quickly seeking to wanted mail inside the file, but it can't compress mails as well * one file at a time compresses better, but seeking is slow because it can only be done by uncompressing all the data until the wanted offset is reached I did a quick test for this with 27 MB of my old INBOX mails: (note the -b option, so it doesn't count wasted fs space) mdbox/storage% du -sb . 15120350 . Maildir/cur% du -sb . 16517320 . % echo 1-15120350/16517320|bc -l .08457606924125705623 So, compressed mdboxes take 8.5% less space. This was with regular gzip compression with default level. With bzip2 -9 compression the difference was 10%. Any thoughts on if 8-10% is significant enough improvement to make seeking performance worse? Or perhaps I should just implement both ways.. :)

4 6