Postfix and 9.5.1: Sending not possible / delayed when Greylisting switched on

[openairhoster2]

Hi,
after updating to Plesk 9.5.1 (Debian Lenny 64), our users are getting
"451 mail server temporary failed" when sending out a mail. After a few minutes, they can send mail to that address. But if they send mail to another address, again, they get that error, have to wait for a few minutes, and then can send mail to that address.
As if greylisting was interfering with sending mail.
And, yes, if we switch off Greylisting, sending works without delay.
This is, of course, not a nice choice: either not being able to send right away, or drown in spam
Replacing postfix-queue with on of the zillion hotfixes of the last months didn't help
Please, please, take a look and fix this

Below an excerpt form the maillog. Hostname and IP has been altered for privacy reasons. The mail client used was Pine from the same server as the mail server. But it doesn't work from other servers, either.

####

Apr 25 18:09:55 mail postfix/smtpd[24330]: connect from mail.xanaluca.com[148.198.112.51]
Apr 25 18:09:55 mail pop3d-ssl: Connection, ip=[95.117.255.168]
Apr 25 18:09:55 mail postfix/smtpd[24324]: connect from localhost.localdomain[127.0.0.1]
Apr 25 18:09:55 mail postfix/smtpd[24330]: NOQUEUE: client=mail.xanaluca.com[148.198.112.51]
Apr 25 18:09:55 mail postfix/smtpd[24324]: B896B480227D: client=mail.xanaluca.com[148.198.112.51]
Apr 25 18:09:55 mail before-queue[24340]: check handlers for addr: [email protected]
Apr 25 18:09:55 mail before-queue[24340]: check handlers for addr: [email protected]
Apr 25 18:09:55 mail before-queue[24340]: Processing handlers...
Apr 25 18:09:56 mail before-queue[24340]: hook_dir = '/opt/psa/handlers/before-queue'
Apr 25 18:09:56 mail before-queue[24340]: call_handlers: call executable = '/opt/psa/handlers/info/05-grey-ZiyzFF/executable'
Apr 25 18:09:56 mail greylisting filter[24346]: Starting greylisting filter...
Apr 25 18:09:56 mail postfix/smtpd[24329]: table hash:/var/spool/postfix/plesk/poplock(0,lock|fold_fix) has changed -- restarting
Apr 25 18:09:56 mail postfix/smtpd[24334]: table hash:/var/spool/postfix/plesk/poplock(0,lock|fold_fix) has changed -- restarting
Apr 25 18:09:56 mail postfix/smtpd[24321]: table hash:/var/spool/postfix/plesk/poplock(0,lock|fold_fix) has changed -- restarting
Apr 25 18:09:56 mail postfix/smtpd[24289]: table hash:/var/spool/postfix/plesk/poplock(0,lock|fold_fix) has changed -- restarting
Apr 25 18:09:56 mail before-queue[24340]: handlers_stderr: DEFER
Apr 25 18:09:56 mail before-queue[24340]: call_handlers: DEFER during call '/opt/psa/handlers/info/05-grey-ZiyzFF/executable' handler
Apr 25 18:09:56 mail before-queue[24340]: call_handlers: stop call handlers from dir '/opt/psa/handlers/before-queue/global'
Apr 25 18:09:56 mail postfix/smtpd[24324]: disconnect from localhost.localdomain[127.0.0.1]
Apr 25 18:09:56 mail postfix/smtpd[24330]: disconnect from mail.xanaluca.com[148.198.112.51]
Apr 25 18:09:56 mail postfix/smtpd[24330]: table hash:/var/spool/postfix/plesk/poplock(0,lock|fold_fix) has changed -- restarting

 

[IgorG]

Developers have informed that it's not bug. It's CORRECT working of greylistinfg feature. See also: http://www.greylisting.org for get complete information. Unfortunately we can't to write correct explanation to remote server in SMTP session because it is require full API rewriting.

 

[openairhoster2]

Maybe I was not clear enough about what happened:

1. This happens only since the upgrade to 9.5.1. It worked in 9.3 and before (except of course for the other Plesk-Postfix bugs that we all know)

2. When I send out mail from "[email protected]" to "[email protected]", my mail server (for example mail.mypleskdomain.com) should get a defer message from Yahoo, if Yahoo uses greylisting. This message should never appear to me, the client.
But it is not Yahoo who sends the defer message. It is mail.mypleskdomain.com sending the defer message directly to the me, the client, which is wrong. So it seems that Plesk/Postfix is doing some kind of greylisting in the wrong place.

3. I don't understand the sentence "Unfortunately we can't to write correct explanation to remote server in SMTP session because it is require full API rewriting.". Maybe you can reword it? But I don't think solving this problem requires full API rewriting, since it worked in 9.3.

Greetings
Michael

 

[IgorG]

Did you tried to rebuild mailsettings with 'mchk --with-spam' ?

 

[openairhoster2]

No, I haven't tried that. Will do so tonight, because it takes too long now, the mail server downtime would drive our customers crazy.

 

[andreyb]

The same problem http://forum.parallels.com/showthread.php?t=100955

 

[charliem]

Igor,

This is the exact same issue I asked you about here...

http://forum.parallels.com/showthread.php?p=411418#post411418

Why could you not respond to my request with helpful advice.

Q. What will "mchk --with-spam" do?

Q.The OP'er seems to think "mchk --with-spam" will cause downtime, how much downtime?

I do hope you can respond to me this time, or would you prefer to ask me for "proof" that I also have this issue again before you are prepared to help me?

 

[DariuszR]

Same problem here after upgrade to 9.5.1 on CentOS 5.4. Some maillog messages:

Apr 26 09:46:41 xxx before-queue[17423]: check handlers for addr: [email protected] user on my server trying to send email
Apr 26 09:46:41 xxx before-queue[17423]: check handlers for addr: [email protected] remote email address
Apr 26 09:46:41 xxx before-queue[17423]: Processing handlers...
Apr 26 09:46:42 xxx before-queue[17423]: hook_dir = '/usr/local/psa/handlers/before-queue'
Apr 26 09:46:42 xxx before-queue[17423]: call_handlers: call executable = '/usr/local/psa/handlers/info/05-grey-Tw58FY/executable'
Apr 26 09:46:42 xxx greylisting filter[17428]: Starting greylisting filter...
Apr 26 09:46:42 xxx before-queue[17423]: handlers_stderr: DEFER
Apr 26 09:46:42 xxx before-queue[17423]: call_handlers: DEFER during call '/usr/local/psa/handlers/info/05-grey-Tw58FY/executable' handler
Apr 26 09:46:42 xxx before-queue[17423]: call_handlers: stop call handlers from dir '/usr/local/psa/handlers/before-queue/global'

 

[openairhoster2]

Unfortunately, mchk --with-spam didn't help ...

 

[IgorG]

Guys, thank you for provided logs and details. I have forwarded it to developers for the further investigation. I will update this thread with results.

 

[openairhoster2]

Hi Charlie,

I was advised to stop mail service during mchk --with-spam, that's the downtime. On the server I am talking about, this take about 20 minutes. But of course this can depend on server hardware, number of mailboxes, etc. Maybe some configuration, too. I have no experience with this.

But as noted, mchk --with-spam doesn't help here, anyway

Igor,

Q.The OP'er seems to think "mchk --with-spam" will cause downtime, how much downtime?

 

[charliem]

Hi Charlie,

I was advised to stop mail service during mchk --with-spam, that's the downtime. On the server I am talking about, this take about 20 minutes. But of course this can depend on server hardware, number of mailboxes, etc. Maybe some configuration, too. I have no experience with this.

But as noted, mchk --with-spam doesn't help here, anyway

Thanks,

Nice that someone has tried to help.

Unfortunately I have been forced to start the process of having cpanel installed on all my servers as there is no way Pleask is tenable now. You should see my thread, it took 6 days for Igor to even tell me what he needed in order to help me. Unfortunately he is a sarcastic, unhelpful individual with poor communication skills. Faced with that, the fact that the only route to get help is to pay for it and the fact that the software has not worked in an even slightly usable manor in 3 upgrades has forced my hand here.

Thanks anyway!

 

[openairhoster2]

So ... is there any solution to this problem?

 

[oscarf]

Same problem here after upgrade to 9.5.1 on CentOS
The problem it`s greylist
call_handlers: DEFER
call_handlers: DEFER

Solution?

 

[karesmakro]

Same problem to me using plesk 9.5.1 on debian etch 64bit!
Trying to send an email to locally existing mal account leads to a deffered/greylisted message, which should not ever happens!
Sending to non locally existing mail accounts is working correct!

I make thoughts about to plan legal actions, to get my money back for licences and for my customers I lost, all my lost time I spent for getting a working system, all my frustration and anger, which stops me to make useful work and earn money. I'm also a provider, developper and supporter for open source software, but never saw anything like this!

You can believe me, I'm more than frustrated and my next step is to migrate to an other system, as I posted in another thread earlier, if I find time!

- postfix queue file write error
- greylisting failes
- horde failes

what is your next surprise?

I love plesk :ironic:

 

[nixxnutz]

Same problem here after upgrade to 9.5.1 on Debian Lenny 5.0

 

[karesmakro]

After some experiments,
I used postfix-queue file from Dec 10. 2009 (I think, it's from Plesk 9.2.3) and all went o.k. until now!

-rwxr-xr-x 1 popuser root 89K 2009-12-10 07:27 postfix-queue

since 16h no greylisting problems and no queue file write error so far

whether postfix-queue file from 9.5.1, nor 9.3, nor Plesk patch file was working

 

[ISSchuster]

Same problem here - Debian 5 / 32bit...

@IgorG: Seems that Parallels is not really interested in solving this problems... It's unbelievable that developers are not able to provide a fix within one week...

 

[openairhoster2]

We are all hostages of Parallels.

Same problem here - Debian 5 / 32bit...

@IgorG: Seems that Parallels is not really interested in solving this problems... It's unbelievable that developers are not able to provide a fix within one week...

 

[openairhoster2]

karesmakro,
do you know how one can the postfix-queue from 9.2.3.? I can download the plesk package 9.2.3, but it contains a lot of directories and debian packages. In which of them would I find postfix-queue?

After some experiments,
I used postfix-queue file from Dec 10. 2009 (I think, it's from Plesk 9.2.3) and all went o.k. until now!

-rwxr-xr-x 1 popuser root 89K 2009-12-10 07:27 postfix-queue

since 16h no greylisting problems and no queue file write error so far

whether postfix-queue file from 9.5.1, nor 9.3, nor Plesk patch file was working