01-12-2015 07:19 AM
My infrastructure team wants to move our 2 clustered Ironport C370's to new IP spaces, each at separate data centers. Something I've never done before. What pages do I need to change in the config before they physically move the boxes to the new datacenter in a new subnet? What firewall rules need to be allowed?
Thanks.
Solved! Go to Solution.
01-12-2015 07:48 AM
Hi,
change IP addresses for all interfaces you need to: Network > IP interfaces.
Change IP routes: Network > routes
If your mail servers (Exchange) gets new IP addresses also then you should change SMTP routes also: Network > SMTP routes
Are there new DNS servers? Network > DNS
If LDAP's IP is changed: System administration > LDAP
Syslog server if needed.
Firewall rules:
1. incoming mail: at least port tcp/25 in both directions (check Network > Listeners if there are any other ports configured for listeners).
2. management: usually tcp/443 for GUI and tcp/22 (CLI-ssh). If you're still using telnet then open port tcp/23 also.
3. SPAM quarantine: tcp/82 (http) and tcp/83 (https) usualy; if users are authentication via POP/IMAP then open tcp/110/143 also.
4. ftp (if needed) for file transfer (in or/and out): logs, configuration...
5. DNS (outgoing): tcp+udp/53
6. syslog if it's used (outgoing): udp/514
7. if you're using AD/LDAP integration: tcp/389 or tcp/636 (for secured connection)
8. HTTPS (outgoing) for updates
As I understand you have two C370 in cluster and you're not using SMA. If you're using SMA also then you need to open additional ports (tcp/6025 and tcp/7025).
I hope I didn't forget some :)
01-12-2015 07:48 AM
Hi,
change IP addresses for all interfaces you need to: Network > IP interfaces.
Change IP routes: Network > routes
If your mail servers (Exchange) gets new IP addresses also then you should change SMTP routes also: Network > SMTP routes
Are there new DNS servers? Network > DNS
If LDAP's IP is changed: System administration > LDAP
Syslog server if needed.
Firewall rules:
1. incoming mail: at least port tcp/25 in both directions (check Network > Listeners if there are any other ports configured for listeners).
2. management: usually tcp/443 for GUI and tcp/22 (CLI-ssh). If you're still using telnet then open port tcp/23 also.
3. SPAM quarantine: tcp/82 (http) and tcp/83 (https) usualy; if users are authentication via POP/IMAP then open tcp/110/143 also.
4. ftp (if needed) for file transfer (in or/and out): logs, configuration...
5. DNS (outgoing): tcp+udp/53
6. syslog if it's used (outgoing): udp/514
7. if you're using AD/LDAP integration: tcp/389 or tcp/636 (for secured connection)
8. HTTPS (outgoing) for updates
As I understand you have two C370 in cluster and you're not using SMA. If you're using SMA also then you need to open additional ports (tcp/6025 and tcp/7025).
I hope I didn't forget some :)
01-16-2015 09:36 AM
This was a huge help.
Once I change the IP's of the Network Interfaces I assume I'll lose connectivity to the equipment right?
So at that point I can just power them off and move to the new data center?
Thanks
01-16-2015 09:28 PM
Yes, you'll lose connectivity to the ESA after you commit changes (assuming you're changing management interface's IP address).
Shutdown gracefully the appliance if possible before unplugging the power cord (using serial interface, by plugging laptop directly to ESA with temporary assigned IP address from a scope of newly assigned subnet range...).
If you're using IP addresses for cluster communication you'll have to update configuration as Matthew pointed out. I'd recommend you to set communicate by hostname to avoid repeating this procedure next time you'll changing IP addresses of appliances. You'll only have to do change IP address in one place only - DNS server.
01-15-2015 10:43 AM
You will probably need to re-establish the cluster. Both boxes are probably going to report that they can't find the cluster peer. You might be able to do it like this:
Machine B - tear down cluster and join to existing cluster on A
Machine A - tear down cluster and join back to B
That would get the cluster back up and running. However, if you don't have a tremendous amount of machine-level exceptions you might be better off just tearing down the whole cluster and starting from scratch so it will be a "clean" config.
01-15-2015 06:36 PM
You could also just follow through with the IP changes on the interface configuration as Jernej has pointed out
And if the cluster decides to complain that it cannot reach the cluster peer due to network communication error (the likely error in the event IP changes) you can check the clusterconfig > communication > and update the IPs if it is not updated, this would clear the cluster peer communication problems and they'll both sync back up to normal.
This is to avoid breaking down the cluster :)
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide