451 Error: queue file write error (only when using postfix)

[fearz]

Guys, Any updates?

I have the same issue,

Kindly advise,

Thanks

 

[albans]

Same issue after installing this hotfix: http://kb.odin.com/en/6074

Now, when the error happens, the log in maillog is as follow:
Mar 5 06:06:25 myhost postfix/spawn[926]: warning: command /usr/lib/plesk-9.0/postfix-queue killed by signal 11
Mar 5 05:06:25 myhost postfix/smtpd[891]: warning: lost connection with proxy 127.0.0.1:10025
Mar 5 06:06:25 myhost before-remote[929]: Lost connection
Mar 5 06:06:25 myhost before-remote[929]: Some error occured
Mar 5 05:06:25 myhost postfix/smtpd[930]: lost connection after DATA from unknown[127.0.0.1]
Mar 5 05:06:25 myhost postfix/smtpd[930]: disconnect from unknown[127.0.0.1]

I'm still looking for a solution!

 

[Yves Vogl]

Hi,

we've switched to qmail by using the auto installer. This solved the problem.
Maintaining the postfix course was beyond the pale after contacting the platinum reseller support (through United Internet AG) 3 times within 2 weeks without having the problem solved.



It seems that postfix support in Plesk 9 has some smell of a bluff package*– starting with placing the files within the /var/qmail dir and ending with several serious bugs because of the propietary binaries interspersed to tie the customer to Plesk.

 

[Markus_Eyrich]

I can't understand this, it seem's parallels doesn't care about this at all. A normal Postfix Mail Server is set up in 20 Minutes with full featured spam filters, virus filters etc and working like a charm. Postfix problems are there since 9.0 and they don't fix it. Queue File write errors, turning on dr. web does even more nonsense, no rejecting of non-existent mailboxes on alias domains and so on and so on....i'm really asking myself why my organisation payed money for a product that isn't working at all...can't be that hard to fix, and for god's sake, those erros are showstoppers, bugs like that need to get fixed YESTERDAY not with the next update or anything like that. If we would do this with our products, we would have no customers at all...

 

[Yves Vogl]

Full ACK – and last but not least it shows that the Parallels Platinum Support is completely worthless

 

[105547111]

The kill signals I found cause the postfix-queue to then segfault - disgraceful of needs to exit properly not just freeze in it's process.

Any program segfaults it's a serious problem.

Still no fix!

I feel sorry for you. I paid for a bloody ticket just for postfix issues. Then they just say fixes will be coming at some stage closed the ticket. But it says you can reopen it, how it's closed there is no reopen. I then try replying to the last email I get one back it's closed you need to purchase another!

What the hell!

What is worse during the ticket another issue raised itself (probably with the patches who knows) so I bought another ticket.

Today 2 tickets issued and still not fixed I refuse to buy any more!

I sent a PM to the admin giving my ticket number asking it to be reopened - nothing

How is this fair and a way to treat your customers?

 

[Markus_Eyrich]

As we can see, still no answer from an parallels representative about the situation....really sad..

 

[lordshinva]

May be a problem related to selinux?
What is your selinux state?

 

[Yves Vogl]

Maybe someone should give c't (Heise) a hint to read this thread. It would not be the first time that publicity acts as a motivation for managers to get off their seats

 

[Yves Vogl]

My previous post where I suggested informing some guys from a german IT magazine was killed by censorship. Maybe we should get it up in the blogosphere then, instead?

 

[105547111]

It's not selinux, I already ran in debug mode to make sure no audits are hidden and there is nothing postfix anything being denied.

It's just plainly crashing as it's stopping due to this processing failure and hence segfaulting. It's clearly the postfix-queue that is segfaulting and failing.

I wasted enough already despite a ticket it's still doing this

 

[fearz]

That is really redicilous Plesk Employees...
That is a serious application that is supposed to be completely reliable and stable. its been a month+ now with that issue bieng postponed, i'm coming here everyday to check for any updates, and its extremly frustrating that you don't even reply with an official word to your uses.
Your application is not free, your support is not free yet we are getting nothing in return.

ATLEAST GET BACK TO US WITH ANYTHING.

 

[Markus_Eyrich]

Well, either they don't care or got told not to answer. I now informed some members of heise.de aswell as other news sites which regulary post updates on products like plesk, let's see if some of them get here to the forum reading about all the mess and maybe write some real nice (NOT) articles about such a lousy product and customer support. I also informed my representatives that i made a mistake by choosing plesk for our customer servers, thanks for that parallels, since i'm now the 'idiot' in my company which voted for plesk....

At current state, plesk is not worth a cent...

 

[PSi_101]

I too am having this problem.
Has anyone tried experimenting with the permissions and ownership of the postfix-queue file?
I noticed that in the KB article it says to set the ownership and perms to mhandlers-useropuser, 550. And goes on to give the example commands for CentOS.
Does this apply to Debian too?
I'm running Debian 4

 

[galaxy]

The "hotfix" for postfix at;

http://kb.odin.com/en/6074

is a replacement "postfix-queue" executable to increase the timeout from 30 seconds
to 300 seconds and allows plesk to process attachments up to 10M on a medium
loaded machine.

There are two problems here. I saw in a prior post that the user gets signal 11, which
is a segmentation violation, which translates into a programmer bug/error. Parallels needs
to fix it and post an update. Second, the compiled in value for the timeout is now
300 seconds which should not be compiled in. It should be configurable. We allow
customers 30M attachments, which would imply 900 seconds for us, but there are peak
times when the machine is heavily loaded, so even the 900 may not be acceptable.

So apparently they know the problem (the postfix-queue executable), however they
posted a buggy fix. And the fix isn't useful for folks that still want to allow large attachments.

It would be better if they can post the source code and we can do what we want with it.
Otherwise, they need to make it configurable. And this needs to come out NOW, as a
hotfix since there's so many broken installations already.

 

[albans]

The following workaround did not work for me:
"For us it worked to switch Dr.Web OFF, SPF OFF, Domainkeys OFF, DNS Blachhole ON. Then restart mail service. "
Parallels: do you have a solution??

 

[105547111]

They keep bloody changing this hotfix. The files keep changing sizes as well as timestamps. How about a revision history or change list?

 

[galaxy]

Probably to fix the bug that caused the SEGV. The solution still isn't acceptable though. It gets you to 10M attachments.

What exactly does postfix-queue do? Does anyone specifically know? I would like to remove that totally from master.cf since it is a bottleneck (program fork/exec for every inbound and outbound message).

Can I safely eliminate that program if I don't use Dr. Web or Spamassassin?

It would be ideal if I could remove it, or if its needed I'd rather have the source code so I can streamline the process into a daemon rather than a program.

 

[105547111]

Please keep us updated on hotfixes that are updated.

Sad enough plesk 9 is full of bugs but please pay attention to provide correct hotfixes please!

Also can't all these be pushed though as plesk updates, how many suffer and unless they recheck the dates think they are already updated, not to mention other customers unaware of the hotfix?

If your not willing to push updates through updater, then please a dedicated forum only writable my moderator / admin stating the hotfix and therefore updates are in the post or use the useless empty announcements for such important fixes please!

Edit - Despite this 'new' version it still segfaults yet another failure

 

[PSi_101]

Come on Parallels, we're still waiting for the fix!