Invalid_PDU error (OS 4.5)

perimere · Post by **perimere** » Sat Jul 21, 2007 7:37 pm

Hi All,
First off I'm really impressed with Zimbra. Had it running about a month now and it was a breeze to install, and worked like a charm. I'm really looking forwards to the Tasks in 5.0

Now the bad news .....
I'm experiencing the exact same problems as a previous thread:

I">http://www.zimbra.com/forums/installati ... -ga-4.html
I also followed the steps 5 - 7 as the resolution detailed in the thread, however it doesn't appear to work for me.
Has anyone else encountered this, or have an idea of how I may progress?
TIA
Cheers
Adam

perimere · Post by **perimere** » Sun Jul 22, 2007 5:11 am

OK,
I tried connecting from XP SP2 and IE6, and received a different error:

AjxException.INVALID_PARAM.
The same issue is present in the basic client as previously reported.
Trying to connect to the Administration Console is still blanks, however IE reports errors on the page. The details of which reveal:

Line: 22301

Char: 1

Error: Exception thrown and not caught.
Are there any log files that may shed some light on this? I had a quick look last night (2am *sighs*) and didn't see anything useful, but just means I don't know what I was looking for

FYI, the box is a Debian 4.0 machine 64-bit HW, but running 32-bit OS. 2GB RAM, and dedicated to Zimbra - there's nothing else putting a load on the box!
This is my home mail server with 1GB of mail stored for the family, and hosts 6 mail domains so I'd really like to sort this ASAP.
TIA
Cheers
Adam

bobby · Post by **bobby** » Tue Jul 24, 2007 12:42 pm

What is the output of "zmcontrol status"?

perimere · Post by **perimere** » Tue Jul 24, 2007 5:42 pm

Hi Bobby,
I appreciate the help here. I've had to roll back to the old mail server until I can resolve this

The output is:

[QUOTE]

antispam Running

antivirus Running

ldap Running

logger Stopped

zmlogswatchctl is not running

mailbox Running

mta Running

snmp Running

spell Running

[/QUOTE]

Running zmcontrol start returns:

[QUOTE]

Starting logger...Done.

Starting mailbox...Done.

Starting antispam...Done.

Starting antivirus...Done.

Starting snmp...FAILED

Starting spell...Done.

Starting mta...Done.

[/QUOTE]

Checking with zmcontrol status:

[QUOTE]

antispam Running

antivirus Running

ldap Running

logger Running

mailbox Running

mta Running

snmp Running

spell Running

[/QUOTE]

However the symptoms of being unable to login and no admin console persist.
It's just a thought, but I'm concerned that Debian has a software update mechanism similiar to Windows Update. I'm wondering if a dependent component has been upgraded breaking the installation?
Looking at available updates for the system, I'm showing libcurl3 (7.15.5-1etch1) for example. I'm wondering what sort of impact this would have on the Zimbra installation?
Have any other Debian users experienced issues with components breaking the installation?
Any suggestions on how to progress further?
TIA
Cheers
Adam

bobby · Post by **bobby** » Wed Jul 25, 2007 7:06 pm

It is possible. If there is a problem with the updated libcurl, we should see errors in the logs. Are there stacktraces in /opt/zimbra/log/mailbox.log that correspond to the failed login attempts?

perimere · Post by **perimere** » Thu Jul 26, 2007 12:47 am

Hi Bobby,
I'll check shortly. I haven't installed the libcurl update. It was just that I saw it pending and realized that this is the only feasible explanation of why the install suddenly stopped working (i.e. other components have been updated through this mechanism).
Oh, the only other software on the box is Bind9 configured for split brain (internal / external views).
Cheers
Adam

perimere · Post by **perimere** » Thu Jul 26, 2007 1:59 am

Hi Bobby,
I checked the logs as mentioned. I can see errors complaining of being unable to bind to the external IP address of the MX record. This isn't present in logs previous to last Friday (the day the problem commenced).
What's interesting here is that the nameserver being used by the box is local, and resolves to a reserved internal IP address.
I ran the zmdumpenv script and there is no sign of the external IP anywhere. It does however identify the internal IP of the server.
I can appreciate I may have configured the server incorrectly, maybe I'm not supposed to put the server behind NAT, with internal/external IP's? Though I've never had that problem with other mail products in the past.
I feel like we're closing in on the problem though

Cheers
Adam

perimere · Post by **perimere** » Thu Aug 02, 2007 9:49 am

Just in case anyone else has this problem, I'll detail the solution to how to fix this here. I still don't know what caused this, but Bobby was excellent inhelping walk this through! Thanks.
The mailbox.log file indicated that the tomcat server was trying to bind to my external IP address (NAT), and not the internal IP address it was supposed to.
DNS, and all the obvious configuration appeared to point to everything being set to the local IP address.
Howerver, the following identified the culprit:

zmprov --ldap getServer MyServer.domain.tld | grep Lmtp
[QUOTE]

zimbraLmtpAdvertisedName: mail.domain.tld

zimbraLmtpBindAddress: 82.xxx.xxx.xxx

zimbraLmtpBindOnStartup: FALSE

zimbraLmtpBindPort: 7025

zimbraLmtpNumThreads: 20

[/QUOTE]
Set back to the correct, internal IP, with:

zmprov --ldap modifyServer MyServer.Domain.tld zimbraLmtpBindAddress

192.168.xxx.xxx
No idea, how/why it changed, as I had made no changes since install, but it's fine now.
Thanks again guys

Cheers
Adam