Postfix still sending hundreds 'queue file write error'

[charliem]

Guys,

This problem Postfix "queue file write error" is really became priority for us and I would like to receive the following information from you:

1. Version of Plesk and OS. Specify, if it is Virtuozzo container.
2. What of the published Postfix patches have been installed.
3. Do you have enabled any spam protections.
4. Conditions for occurrence of this error that you have noticed.
5. Your any additional information which you considers necessary to inform about this problem.

Please answer strictly on these points, shortly and with necessary details.
On the basis of this collected information I will prepare the analytical report for developers.
Thank you for cooperation.

1 - centos5 64 - Linux 2.6.18-128.el5
2 - this one http://forum.parallels.com/showpost.php?p=396040&postcount=43
3 - have tried with and without
4 - usually attachments over about 4MB
5 - none

 

[MathiasD]

I've got the same problem on CentOS 5, installed Qmail for the time being but want to go back to postfix as soon as patch is available. Do you have any idea when this patch is available?

 

[charliem]

gone quiet

well, this patch was in testing, now you've gone quiet, did I miss something, no post of where the latest patch can be found just a note about 9.5 in April. Am I supposed to tell my customers that can send emails with attachments in April. until then you can't so go find another host? Not good. I can feel a control panel installation coming on.

 

[105547111]

9.5 may be held up a few days due to this OpenSSL issue stopping plesk panel running on Fedora, Redhat and Centos.
I sm sure developers are jumping at that no GUI is critical.

 

[mattclements]

Guys,

This problem Postfix "queue file write error" is really became priority for us and I would like to receive the following information from you:

1. Version of Plesk and OS. Specify, if it is Virtuozzo container.
2. What of the published Postfix patches have been installed.
3. Do you have enabled any spam protections.
4. Conditions for occurrence of this error that you have noticed.
5. Your any additional information which you considers necessary to inform about this problem.

Please answer strictly on these points, shortly and with necessary details.
On the basis of this collected information I will prepare the analytical report for developers.
Thank you for cooperation.

1. Plesk 9.3.0 & CentOS 5.4
2. smtp_proxy_timeout changed to 600s
3. All disabled - Spam Filtering provided by a different solution
4. Mail receieved by our spam filter box, filtered and tries to deliver to our Postfix SMTP, receives "Mar 30 20:33:23 spamwall/smtp[46596]: 60AA5140EB3: to=<[email protected]>, relay=82.165.194.xxx[82.165.194.xxx], delay=174909, status=deferred (host 82.165.194.xxx[82.165.194.xxx] said: 451 4.3.0 Error: queue file write error (in reply to end of DATA command))"
The mail sucessfully delivers to the receipient, but as the response is above, the mail is defered and retried every 10 minutes. So repeats every 10 minutes with the end user getting the email 6 times an hour

 

[mattclements]

Guys Guys - As said above this is in testing and therefore can not be rushed - Imagine fixing this damn mail problem - but the web server not working, or FTP uploads.

However this fixed my problem entirely with the queue file write error - it will remove the spam filtering completly - but we use a seperate box for filtering as not to slow down our web servers too much: http://se.id.au/journal/files/plesk_postfix_queue_file_write_error.php

Hope it helps,
Matt

 

[Flamenetworks]

Hello there,

is it possible to have a release date?

We've many Plesk servers and customers are getting nervous because of this bug.

Regards,

F.

 

[IgorG]

Once more again - WE NEVER PROVIDE ANY ETA AND RELEASE DATES.
I can say only that Plesk 9.5.0 will be released in April.

 

[jacko@]

Hello there,

is it possible to have a release date?

We've many Plesk servers and customers are getting nervous because of this bug.

Regards,

F.

We made the mistake of using the Postfix implementation within Plesk 9 and saw pretty much all the errors discussed in this thread. Yesterday we had to change back over to qmail to resolve the issues. Surprisingly when the migration over to qmail completed it appears to have worked without issue.

Seems that my previous post was deemed unacceptable regarding upgrading the 9.5 when it comes out, however, wishing you luck when you choose to update to it (obviously once released).

jacko

 

[CatalinM]

I do have the last patch for postfix-queue but as others said the 'timeout talking to proxy' is still present from time to time ; the below tests are done to a VPS CO5.4 32b running latest plesk 9.3 . What I had found is .. the error is related to the size of the message + size of hostname + size of user mail rounded to possible 4096 . More details
I have 2 smtp session files which differ by only 1 byte
diff issue-plesk-postfix.txt issue-plesk-postfix-2.txt

< <fead>
---
> <ead>

one hits the bug , the other is not

$ nc 1.1.1.1 25 < issue-plesk-postfix-2.txt
220 343243243242.xx ESMTP Postfix
250-343243243242.xx
250-PIPELINING
250-SIZE 52428800
250-VRFY
250-ETRN
250-STARTTLS
250-AUTH DIGEST-MD5 PLAIN CRAM-MD5 LOGIN
250-ENHANCEDSTATUSCODES
250-8BITMIME
250 DSN
250 2.1.0 Ok
250 2.1.5 Ok
354 End data with <CR><LF>.<CR><LF>
250 2.0.0 Ok: queued as BE86FF1A836

$ nc 1.1.1.1 25 < issue-plesk-postfix.txt
220 343243243242.xx ESMTP Postfix
250-343243243242.xx
250-PIPELINING
250-SIZE 52428800
250-VRFY
250-ETRN
250-STARTTLS
250-AUTH DIGEST-MD5 PLAIN CRAM-MD5 LOGIN
250-ENHANCEDSTATUSCODES
250-8BITMIME
250 DSN
250 2.1.0 Ok
250 2.1.5 Ok
354 End data with <CR><LF>.<CR><LF>
451 4.3.0 Error: queue file write error

I think this is related to the fact that /usr/lib/plesk-9.0/postfix-queue is not seeing the end of smtp session '.\r\n'
when the special condition occurs
Hopefully this will be enough to fix this bug

 

[IgorG]

http://kb.odin.com/en/8391

 

[cgmm]

Igor, this have nothing to do with that KB , the error I have is 100% reproductible due postfix-queue .
If you can not pass this to devels at least tell me what more info you need and I will provide it.

 

[cgmm]

Try the below perl script a few times (put it in a plesk server ):
it needs nc (in CO5 : yum install nc)

You will see that sometimes it hangs, while others is not.. but mail is delivered each time. Each time it hangs
will result in a timeout proxy error and duplicate delivery

perl ./aaa.pl `postconf -h myhostname` | nc 0 25



#!/usr/bin/perl


print 'EHLO localhost
MAIL FROM: <[email protected]>
RCPT TO:<[email protected]>
DATA' . "\n\n";

for (1..97) { print 1 x 80 . "\n" }

print 1 x (132 - length $ARGV[0]) . "\n";
print "." . "\n";

 

[IgorG]

cgmm,
Thank you for cooperation. I have forwarded it to developers for further investigation.

 

[IgorG]

cgmm,

This example shouldn't work. Because non RFC end-of-line was used.

print 'EHLO localhost
MAIL FROM: <[email protected]>
RCPT TO:<[email protected]>
DATA' . "\n\n";

for (1..97) { print 1 x 80 . "\n" }

print 1 x (132 - length $ARGV[0]) . "\n"; print "." . "\n";

 

[cgmm]

Igor,
Try the below updated script like this

plesk_server# ./make-postfix-plesk-issue.pl `postconf -h myhostname` 0 | nc 127.0.0.1 25
plesk_server# ./make-postfix-plesk-issue.pl `postconf -h myhostname` 1 | nc 127.0.0.1 25
plesk_server# ./make-postfix-plesk-issue.pl `postconf -h myhostname` 2 | nc 127.0.0.1 25
plesk_server# ./make-postfix-plesk-issue.pl `postconf -h myhostname` 10 | nc 127.0.0.1 25

with 0 and 1 it will timeout (sometimes may not ), but from 2 3 .. up it will work fine (or at least not that offten).


#!/usr/bin/perl

$crlf = "\r\n";

print 'EHLO localhost' . $crlf;
print 'MAIL FROM: <[email protected]>' . $crlf;
print 'RCPT TO:<[email protected]>' . $crlf;
print 'DATA' . $crlf;
print $crlf;

for (1..97) { print 1 x 79 . $crlf }

print 1 x (145 - length($ARGV[0]) + $ARGV[1]) . $crlf;
print "." . $crlf;

 

[IgorG]

cgmm,

From developers:

Thanks a lot for you script. Yes. I see little timeout for '0' and '1' numbers.

1. But this timeout is not so often. (actually i saw timeout for a couple(3-4) mails in begin)
2. This not broke the mail transfer.

 

[cgmm]

Igor, indeed that is delivering but is the remote smtp gets a 4xx error which result in a
duplicate deliveries and also bounces sent to local postmaster.
I hope devels are going to fix postfix-queue soon

 

[IgorG]

cgmm,

It shouldn't be due to delay in fractions of a second. Try to enable debug_peer_list (http://www.postfix.org/DEBUG_README.html#logging) and check what is going on there. It would be desirable to receive log which on time coincides with session from the postmaster's email.

 

[cgmm]

Igor,
This is not 'fractions' of second. Is hanging there till it timeouts due smtpd_timeout / smtpd_proxy_timeout .

# time ./make-postfix-plesk-issue.pl `postconf -h myhostname` 0 | nc 127.0.0.1 25
220 xxx1.com ESMTP Postfix
250-xxx1.com
250-PIPELINING
250-SIZE 52428800
250-VRFY
250-ETRN
250-STARTTLS
250-AUTH DIGEST-MD5 PLAIN CRAM-MD5 LOGIN
250-XFORWARD NAME ADDR PROTO HELO SOURCE
250-ENHANCEDSTATUSCODES
250-8BITMIME
250 DSN
250 2.1.0 Ok
250 2.1.5 Ok
354 End data with <CR><LF>.<CR><LF>
451 4.3.0 Error: queue file write error

real 1m40.216s

in my VPS smtpd_proxy_timeout = 100s
Please do not tell me I need to increase that but have devels fix postfix-queue , since that is what is causing all this