[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: mail queue's, ext3 data=journal and sync-mount

On Mon, 19 Aug 2002 00:51, you wrote:
> reply to my last email?  I'm sorry to bother, but I'm trying to find a

I missed that, but it seems you raise the same issues here.

> suitable filesystem and mount options for a qmail queue.  DJB says no to
> ext2 unless it's sync mounted, but I found abysmal performance this way.

Most people find that they can get enough reliability from ext2 without 
synchronous mounting.  People like DJB say "oh the horror, you might lose a 
single email some time", but most admins don't care too much.  As a general 
rule the larger the ISP the more cavalier they are about losing email.  An 
ISP running 1M accounts probably won't be bothered about losing the data from 
10,000 accounts on occasion...

Of course we don't want to be like that, but losing an occasional email in an 
fsck isn't such a big deal.

There is also the outside possibility that a power failure on a busy drive 
could leave the file system so stuffed that FSCK does no good.  I've seen a 
lot of Linux machines treated badly, with power failures, flakey hardware and 
kernel panics, I've only had to use debugfs on a few occasions (and they were 
due to a combination of a kernel bug and a fsck bug - which has since been 
fixed).  Also the debugfs was for a cosmetic thing (if I didn't mind a few 
dud files in lost+found I could have just ignored it).  I've never lost a 
file system through ext2.

But anyway ext2 is old now so don't use it.

Ext3 with data=journal should be better than a synchronous mount in terms of 
data reliability as far as I understand it.  If you have a synchronous 
mounted file system and you write 8K of data then if the write succeeds then 
it's all on disk.  But if the system reboots in the middle then what happened?
Did the file get extended but have no data written?  Did 4K of the 8K get 
written because the two allocation blocks were at different ends of the disk? 
Data journalling should journal both the meta-data and the file data at the 
same time, so the operations of extending the file (meta-data change) and 
that of writing two blocks of data will all be in the same transaction which 
will be atomic.

Of course a mail server will perform a variety of tricks so that these things 
shouldn't be an issue and the features of data journalling should not be 
required.  Writing a mail server would be easier if you could rely on a 
data=journal mounted system!

> Do you know if ext3 with data=journal would be sufficient?

I believe it's more than sufficient, but I'm not a true expert on this.  I'm 
sure someone here will correct me if I'm wrong.

I do not get viruses because I do not use MS software.
If you use Outlook then please do not put my email address in your
address-book so that WHEN you get a virus it won't use my address in the
>From field.

Reply to: