Problem moving ZCS 4.5.9 OSE from FC4 server to FC5 server

Hi,

I am having trouble moving ZCS 4.5.9 OSE from a server running Fedora Core 4 to a new server running Fedora Core 5 (HW not supported under FC4).

I have followed the instructions as described in
http://www.zimbra.com/blog/archives/...er_server.html.

The problem seems to be related to Java and Tomcat. No service is listening on port 7071. From zmsetup.log.X:

Code:

---

COS attribute retrieved for COS default: zimbraFeatureNotebookEnabled=FALSE
*** Running as zimbra user: /opt/zimbra/bin/zmprov mc default zimbraFeatureNotebookEnabled TRUE
[] INFO: I/O exception (java.net.ConnectException) caught when processing request: Connection refused
[] INFO: Retrying request
[] INFO: I/O exception (java.net.ConnectException) caught when processing request: Connection refused
[] INFO: Retrying request
[] INFO: I/O exception (java.net.ConnectException) caught when processing request: Connection refused
[] INFO: Retrying request
ERROR: zclient.IO_ERROR (invoke Connection refused, server: localhost) (cause: java.net.ConnectException Connection refused)
[...] Log-file contiunes with similar messages

---

If I do a clean install on the new server it works. But if I uninstall, do an install.sh -s, merge /opt/zimbra from the old server (tar-ed with tar -cpf), zmfixperms and then install.sh, the problem appears.

Has anybody experienced the same problem, or have any idea what is causing it?

Regards,

Knut

Welcome to the forums.

I assume your new server has the same IP, hostname and DNS A & MX records pointing to it?

Hi,

Yes, the new and the old server has swapped identity. (IP address and hostname.)

Regards,

Knut

I assume you've also disabled all the usual suspects - ipv6, SELinux & firewall? Just for confirmation, can you run the following on the new server:

Code:

---

host `hostname`  <-- type as-is with backticks not single quotes
dig yourdomain.com mx
dig yourdomain.com and

---

and post the output.

Hi,

I have removed IPv6 from /etc/hosts, leaving only the following line:

Code:

---

127.0.0.1              localhost.localdomain localhost

---

IPv6 is disbled in /etc/sysconfig/network:

Code:

---

# grep IPV6 /etc/sysconfig/network
NETWORKING_IPV6=no

---

But it is still on the interface (after reboot):

Code:

---

# ifconfig eth0 | grep inet6
          inet6 addr: fe80::21a:a0ff:fe1b:d70a/64 Scope:Link

---

Is this causing the problem?


SELinux is not enabled:

Code:

---

# selinuxenabled; echo $?
1

---

(From man selinuxenabled: ... status 0 if SELinux is enabled and 1 if it is not enabled.)

Right now the new server is back in the lab, but this is how it looks like on the old (functioning) server (with the correct IP address and hostname):

Code:

---

$ host `hostname`
mail.hjemme.no has address 62.97.193.11
$ dig hjemme.no mx

; <<>> DiG 9.3.1 <<>> hjemme.no mx
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 38945
;; flags: qr aa rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 2, ADDITIONAL: 4

;; QUESTION SECTION:
;hjemme.no.                    IN      MX

;; ANSWER SECTION:
hjemme.no.              3600    IN      MX      3 jesper.bkkb.no.
hjemme.no.              3600    IN      MX      1 barracuda.hjemme.no.

;; AUTHORITY SECTION:
hjemme.no.              3600    IN      NS      ns1.bkkb.no.
hjemme.no.              3600    IN      NS      ns2.bkkb.no.

;; ADDITIONAL SECTION:
barracuda.hjemme.no.    3600    IN      A      62.97.193.12
jesper.bkkb.no.        86400  IN      A      62.97.193.4
ns1.bkkb.no.            86400  IN      A      62.97.193.3
ns2.bkkb.no.            86400  IN      A      62.97.193.53

;; Query time: 0 msec
;; SERVER: 62.97.193.3#53(62.97.193.3)
;; WHEN: Tue Nov 20 15:44:54 2007
;; MSG SIZE  rcvd: 181

kas@mail:~$ dig hjemme.no

; <<>> DiG 9.3.1 <<>> hjemme.no
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 37933
;; flags: qr aa rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 2, ADDITIONAL: 2

;; QUESTION SECTION:
;hjemme.no.                    IN      A

;; ANSWER SECTION:
hjemme.no.              3600    IN      A      62.97.243.54

;; AUTHORITY SECTION:
hjemme.no.              3600    IN      NS      ns2.bkkb.no.
hjemme.no.              3600    IN      NS      ns1.bkkb.no.

;; ADDITIONAL SECTION:
ns1.bkkb.no.            86400  IN      A      62.97.193.3
ns2.bkkb.no.            86400  IN      A      62.97.193.53

;; Query time: 0 msec
;; SERVER: 62.97.193.3#53(62.97.193.3)
;; WHEN: Tue Nov 20 15:44:56 2007
;; MSG SIZE  rcvd: 116

$

---

Swapping the two servers should not change any of this.

As you can see, email is relayed through some anti-spam/anti-virus servers before it hits the Zimbra server.

The servers are running with iptables enabled, and full access for loopback:

Code:

---

-A RH-Firewall-1-INPUT -i lo -j ACCEPT

---

Regards,

Knut

Quote:

---

Originally Posted by kas

Hi,

I have removed IPv6 from /etc/hosts, leaving only the following line:

Code:

---

127.0.0.1              localhost.localdomain localhost

---

---

That's not quite what you want. You should also have a line for your server e.g.:

Code:

---

62.97.193.11 mail.hjemme.no mail

---

For the ipv6 you'll need an extra step. Check these details , you'll also need to reboot (although a network restart may do it).

We don't recommend running a firewall on the Zimbra server itself but If you're going to do it you'll need additional rules in there, check this article.

Hi,

I have now completely disabled IPv6 (and rebooted) and added the requested line to /etc/hosts.

But I do not think that the problem is related to the basic network configuration, because the old server is in production with the ZCS-FC4 install and the new server works with a clean ZCS-FC5 install. The problem appears when the new server is installed (-s), /opt/zimbra is copied from the old server (which is functioning with IPv6 on the interface and only localhost in /etc/hosts) to the new and install rerun.

For some reason Tomcat does not start, and that is why you see the "Connection refused" messages in the log.

I have also added a local nameserver, to make the lab-environment reflect the production-environment:

Code:

---

# host `hostname`
mail.hjemme.no has address 62.97.193.17
# host -t mx hjemme.no
hjemme.no mail is handled by 10 mail.hjemme.no.
#

---

Regards,

Knut

OK, well now you've done the basic network configuration we can have a look at the log. What error is there in catalina.out?

Hi,

Now I think we are on to something. This is /opt/zimbra/apache-tomcat-5.5.15/logs/catalina.out:

Code:

---

Error occurred during initialization of VM
The size of the object heap + VM data exceeds the maximum representable size
Error occurred during initialization of VM
The size of the object heap + VM data exceeds the maximum representable size
Error occurred during initialization of VM
The size of the object heap + VM data exceeds the maximum representable size
Error occurred during initialization of VM
The size of the object heap + VM data exceeds the maximum representable size

---

(Did not know where to look until now.)

Regards,

Knut

Hi,

Found the solution! The new server has too much memory (16GB)... Had to reduce the Java Heap. Reduced tomcat_java_heap_memory_percent from 25 to 10 (for testing), and now it works. ;-)

Is localconfig.xml the right place to do this?

Thank you for pointing me in the right direction!

Regards,

Knut