Re: Postfix Queues on a Ram Disk

2 msg	postmap header_checks error
10 msg	mailbox_command (dovecot lda) stopped working, ...
5 msg	mynetworks via LDAP lookup
7 msg	Local recipient table in LDAP problem
2 msg	proxy_read_maps
1 msg	smtpd_restriction_class and it's lookup
8 msg	Message Rejected
12 msg	Customize to filter attachment ?
2 msg	Relay question
14 msg	unnecessary routing?
4 msg	do not use 'reject_maps_rbl'?
9 msg	Fwd: Postfix in OpenVZ.
2 msg	bypass body_checks for local mail?
4 msg	catchall with relay_recipient_maps
4 msg	'Name server failure' is output
3 msg	smtp_sasl_password_maps lookups

Postfix Queues on a Ram Disk
\ Duane Hill (23 Jan 2008)
. \ Victor Duchovni (23 Jan 2008)
. . \ Duane Hill (23 Jan 2008)
. . . \ Mike Delaney (23 Jan 2008)
. . . \ Victor Duchovni (23 Jan 2008)
. . . . \ Duane Hill (23 Jan 2008)
. . . \ Leonardo Rodrigues Magalhães (23 Jan 2008)
. . . . \ Tarek W. (23 Jan 2008)
. . . . . \ Leonardo Rodrigues Magalhães (23 Jan 2008)
. . . \ Jan P. Kessler (23 Jan 2008)
. . . . \ Leonardo Rodrigues Magalhães (23 Jan 2008)
. . . . . \ Jan P. Kessler (23 Jan 2008)
. \ (Wietse Venema) (23 Jan 2008)
. \ Leonardo Rodrigues Magalhães (23 Jan 2008)
. . \ Leonardo Rodrigues Magalhães (23 Jan 2008)
. . . \ Duane Hill (23 Jan 2008)

15 msg	sendmail compared to smtp sending
3 msg	Postfix does not recognize remote emails
2 msg	How to disable SSLv2 on smtpd opportunistic TLS?

Subject:	Re: Postfix Queues on a Ram Disk
Group:	Postfix-users
From:	Leonardo Rodrigues Magalhães
Date:	23 Jan 2008

Tarek W. escreveu:
>
>> It would be different if we were talking about these new flash-based
>> disks/solid-state disks. They are 'memory-based' disks but acts as real
>> disks. If information is writen, it will continue there in the case of
>> power-loss of rebooting.
>>
>
> Solid state hard drives simulate real hard drives, as such, if the
> power loss or crash occurs before information is written to them from
> RAM, mail will be lost.
>
>
As far as i know, postfix is coded to make sure things are really
written on disk (flushed) before saying the message is OK/accepted.

Once entered the queue, postfix is coded to guarantee no message
will be lost in the queues even during system crashes/reboots. Am i
wrong here ?

If we have a problem (power loss/crash) during disk writing, postfix
will not receive OK from disk subsystem and thus will not give the 'OK
message accepted' to the connection. Thus, sender (which can be a client
or another server) will know message was NOT accepted and therefore try
it again.

--

Atenciosamente / Sincerily,
Leonardo Rodrigues
Solutti Tecnologia
http://www.solutti.com.br

Minha armadilha de SPAM, NÃO mandem email
gertrudes
My SPAMTRAP, do not email it