On Thu, 2009-10-15 at 20:16 -0400, Charles Marcus wrote:
> On 10/15/2009, Timo Sirainen (t...@iki.fi) wrote:
> > The default is 2 MB, and whoever sets it larger deserves what they get
> > (for better or worse).
>
> Just for clarification... what happens when a single message is larger
> than 2MB?
On 10/15/2009, Timo Sirainen (t...@iki.fi) wrote:
> The default is 2 MB, and whoever sets it larger deserves what they get
> (for better or worse).
Just for clarification... what happens when a single message is larger
than 2MB? Is it stored just as a single dbox file all by itself?
On Fri, 2009-10-16 at 00:24 +0200, Mikkel wrote:
> > It doesn't matter what the user's full mailbox size is. It matters how
> > many of these max. ~2 MB dbox files have expunged messages. Typically
> > users would expunge only new mails, so typically there would be only a
> > single file that n
Timo Sirainen skrev:
> On Thu, 2009-10-15 at 10:55 +0200, Mikkel wrote:
>> Some users would have mailboxes a several hundred megabytes and having
>> to recreate thousands of these every night because of a single mail
>> getting expunged a day could result in a huge performance hit.
>
> It doesn'
On Thu, 2009-10-15 at 10:55 +0200, Mikkel wrote:
> > In any case if there are expunged messages, files containing them would
> > be recreated (nightly or something). That'll unfragment the files.
> >
>
> It would be nice if this recreation interval (nightly, weekly, monthly?)
> was made tunable.
Timo Sirainen skrev:
On Wed, 2009-10-14 at 23:59 +0200, Mikkel wrote:
In case of mdbox wouldn't you have the very same problem since larger
files may be fragmented all over the disk just like many small files in
a directory might?
I guess this depends on filesystem. But the files would typica
At 12:18 PM 10/14/2009, Charles Marcus wrote:
Not that you need any help, but I would suggest to just focus on making
the conversion tools (mbox/maildir > (m)dbox, old config > new config,
etc) rock solid, and document any incompatibilities really well.
It'd be nice to think you were kidding, b
On Wed, 2009-10-14 at 23:59 +0200, Mikkel wrote:
> > I'm planning on keeping both of them. And it's not necessarily only
> > because of NFS users. Multi-dbox was done mainly because filesystems
> > suck (mailbox gets fragmented all around the disk). Maybe if filesystems
> > in future suck less, sin
Timo Sirainen wrote:
On Wed, 2009-10-14 at 23:52 +0200, Mikkel wrote:
But it should be able to heal itself using the backup files in version
2.0, right?
That's the theory anyway. :)
How often are they created anyway?
Whenever dovecot.index file would normally get recreated, the old one is
Timo Sirainen skrev:
On Wed, 2009-10-14 at 23:04 +0200, Mikkel wrote:
So basically you prefer mdbox but are maintaining dbox because of its
almost lockless design which is better for NFS users?
Do you consider it to be viable having two different dbox formats or are
you planning to keep only
On Wed, 2009-10-14 at 23:52 +0200, Mikkel wrote:
> Timo Sirainen wrote:
> > On Wed, 2009-10-14 at 23:41 +0200, Mikkel wrote:
> >> Timo Sirainen wrote:
> >>> And you've actually been looking at Dovecot's error log? Good if it
> >>> doesn't break, most people seem to complain about random errors.
> >
Timo Sirainen wrote:
On Wed, 2009-10-14 at 23:41 +0200, Mikkel wrote:
Timo Sirainen wrote:
And you've actually been looking at Dovecot's error log? Good if it
doesn't break, most people seem to complain about random errors.
Well, it does complain once in a while but it has never resulted in dat
On Wed, 2009-10-14 at 23:41 +0200, Mikkel wrote:
> Timo Sirainen wrote:
> >
> > And you've actually been looking at Dovecot's error log? Good if it
> > doesn't break, most people seem to complain about random errors.
>
> Well, it does complain once in a while but it has never resulted in data
>
Timo Sirainen wrote:
And you've actually been looking at Dovecot's error log? Good if it
doesn't break, most people seem to complain about random errors.
Well, it does complain once in a while but it has never resulted in data
being lost in any way. But I guess your point is that this might h
On Wed, 2009-10-14 at 23:18 +0200, Mikkel wrote:
> >> I don't think I've tried that one. Earlier on I experimented with
> >> fsync_disable=yes (which made a huge difference by the way) but that was
> >> before I started using mail_nfs_storage=yes and mail_nfs_index=yes
> >>
> >> I would like to t
Timo Sirainen skrev:
On Wed, 2009-10-14 at 21:14 +0200, Mikkel wrote:
It has been my wish to move to dbox for a long time hoping to reduce
the number of writes which is really killing us.
BTW. Have you tried maildir_very_dirty_syncs=yes setting? That should
reduce disk i/o, and I'm really inter
On Wed, 2009-10-14 at 23:04 +0200, Mikkel wrote:
> So basically you prefer mdbox but are maintaining dbox because of its
> almost lockless design which is better for NFS users?
>
> Do you consider it to be viable having two different dbox formats or are
> you planning to keep only one of them in
Timo Sirainen skrev:
> The main difference is that mdbox needs to lock files when saving
messages. That's not especially nice with NFS. Single-dbox currently
locks index files, but it can be made entirely lockless eventually.
There are also some other differences like:
- in mdbox all messages
On Wed, 2009-10-14 at 21:39 +0200, Andrzej Adam Filip wrote:
> Timo Sirainen wrote:
> > [...]
>
> Do you play to support both ways migration tools?
> dbox <-> maildir(s)
convert plugin can already do that. In v2.0 I'm planning on getting rid
of convert plugin and relying on dsync program. It can
On Wed, 2009-10-14 at 21:28 +0200, Mikkel wrote:
> From your comments it appears like dbox and mdbox are quite different
> in many ways. Is mdbox going to replace dbox completely or are you
> expecting to keep both formats?
> My point is: what's going to be the difference between dbox and mbox
On Wed, 2009-10-14 at 21:14 +0200, Mikkel wrote:
> >> It has been my wish to move to dbox for a long time hoping to reduce
> >> the number of writes which is really killing us.
> >
> > BTW. Have you tried maildir_very_dirty_syncs=yes setting? That should
> > reduce disk i/o, and I'm really inter
Timo Sirainen wrote:
> [...]
Do you play to support both ways migration tools?
dbox <-> maildir(s)
--
[pl>en: Andrew] Andrzej Adam Filip : a...@onet.eu
"Unlike most net.puritans, however, I feel that what OTHER consenting computers
do in the privacy of their own phone connections is their own b
Timo Sirainen skrev:
On Oct 14, 2009, at 7:03 AM, Mikkel wrote:
Now the big question is whether multi-dbox and single-dbox are
compatible formats.
Kind of, but not practically.
If a Maildir->dbox migration is made on a system running dovecot v.
1.1, would it then be trivial later changing t
Timo Sirainen skrev:
On Oct 14, 2009, at 7:03 AM, Mikkel wrote:
It has been my wish to move to dbox for a long time hoping to reduce
the number of writes which is really killing us.
BTW. Have you tried maildir_very_dirty_syncs=yes setting? That should
reduce disk i/o, and I'm really interest
On 10/14/2009 12:24 PM, Timo Sirainen wrote:
>>> BTW. Have you tried maildir_very_dirty_syncs=yes setting? That should
>>> reduce disk i/o, and I'm really interested in hearing how much.
>> What are the downsides? Also, I'm guessing maybe there are certain
>> conditions where you definitely don't
On Wed, 2009-10-14 at 12:18 -0400, Charles Marcus wrote:
> On 10/14/2009, Timo Sirainen (t...@iki.fi) wrote:
> > BTW. Have you tried maildir_very_dirty_syncs=yes setting? That should
> > reduce disk i/o, and I'm really interested in hearing how much.
>
> What are the downsides? Also, I'm guessing
On Wed, 2009-10-14 at 12:18 -0400, Charles Marcus wrote:
> > The main problem with it is actually how to make it enough backwards
> > compatible that everyone won't start hating me.
>
> I honestly don't see that ever happening Timo - and please, don't go
> down the same road Microsoft did with Win
On 10/14/2009, Timo Sirainen (t...@iki.fi) wrote:
> BTW. Have you tried maildir_very_dirty_syncs=yes setting? That should
> reduce disk i/o, and I'm really interested in hearing how much.
What are the downsides? Also, I'm guessing maybe there are certain
conditions where you definitely don't want
On 10/14/2009, Timo Sirainen (t...@iki.fi) wrote:
> I'm trying to get v2.0.0 out pretty quickly though. v2.0.beta1 should
> hopefully be out in less than a month.
Great news on how fast 2.0 is coming along! I finally got the go ahead
to convert my biggest client, but he wants to hold off (politica
On Oct 14, 2009, at 7:03 AM, Mikkel wrote:
It has been my wish to move to dbox for a long time hoping to reduce
the number of writes which is really killing us.
BTW. Have you tried maildir_very_dirty_syncs=yes setting? That should
reduce disk i/o, and I'm really interested in hearing how mu
On Oct 14, 2009, at 7:03 AM, Mikkel wrote:
It has been my wish to move to dbox for a long time hoping to reduce
the number of writes which is really killing us.
Now I wonder what may be the best way of doing so. I'm considering
some sort of intermediate migration where the existing Maildir
It has been my wish to move to dbox for a long time hoping to reduce the
number of writes which is really killing us.
Now I wonder what may be the best way of doing so. I'm considering some
sort of intermediate migration where the existing Maildir users are
changed to single-dbox and then late
32 matches
Mail list logo
