SSL Errors after migrating domain

[SacAutos]

I migrated one of my customer domains from a 9.5.4 Linux server to another on 10.3 . Now the secure side of the website is broken. I receive these errors from various browsers:

Chrome:

Error 107 (net::ERR_SSL_PROTOCOL_ERROR): SSL protocol error.

Firefox:

An error occurred during a connection to www.website.com.
SSL received a record that exceeded the maximum permissible length.
(Error code: ssl_error_rx_record_too_long)

I have other sites with SSL that I have migrated that don't have this issue. So far I've tried reinstalling the cert and rebuilding the conf files. Neither fixed the issue. Help!!

 

[EdwardDekker]

I migrated one of my customer domains from a 9.5.4 Linux server to another on 10.3 . Now the secure side of the website is broken. I receive these errors from various browsers:

Chrome:

Error 107 (net::ERR_SSL_PROTOCOL_ERROR): SSL protocol error.

Firefox:

An error occurred during a connection to www.website.com.
SSL received a record that exceeded the maximum permissible length.
(Error code: ssl_error_rx_record_too_long)

I have other sites with SSL that I have migrated that don't have this issue. So far I've tried reinstalling the cert and rebuilding the conf files. Neither fixed the issue. Help!!

Do you have checked the settings like the IP adress of that website.
Maybe also check if you have migrate the users correctly and attach the users again at there website.

 

[SacAutos]

More details

The two sites on my server are both behind a firewall via NAT. The only difference I can find so far is that the working one has a /conf http.include file that ends with httpd_ip_default.include where the improperly functioning one ends with httpd.include

Obviously Plesk must be treating one differently over the other. How can I affect this change? Any ideas out there?

 

[SacAutos]

Further down the rabbit hole

Since the only difference between my working domain with a dedicated IP and my non-working domain with a dedicated IP address seemed to be the presence of this "_ip_default" file, I thought that I'd dig deeper.

In the /usr/local/psa/admin/conf/ directory I've found two files: vhosts_bootstrap.conf and ip_vhosts_bootstrap.conf . Sure enough, the files contain links to my files. I searched the knowledgebase and found an entry that encouraged users with fouled apache configurations to execute "/usr/local/psa/admin/bin/httpdmng --reconfigure-all", which I did. All I got was fresh versions of the conf files for each domain on the server.

What is coming increasingly clear to me is that there's a flag or something set someplace deep within a database table somewhere inside of Plesk that tells it that I have a domain that uses a dedicated IP address. I've tried a number of tricks so far trying to fool the software into doing that for me, such as moving the domain to another customer under a different IP address and then switching it back. Nothing has worked so far.

Until I get this fixed my customer is out of business because he can't do any SSL e-commerce. Please help!

 

[SacAutos]

Final resolution

Since this was technically a migration issue from 9.5.4 to 10, I took advantage of the free support offer on these problems. A tech from Parallels gleefully logged into my server and fixed the problem. However, shortly thereafter some kind of loop somewhere must have been introduced, as it took my server down. After rebooting several times and monitoring what was going on, I figured out that Apache was the culprit. Since the only and most change was the enabling of my ecommerce domain, I disabled it and restarted Apache. Everything went back to normal after that.

Fortunately I have another server also running 10.3 available so I migrated the disabled domain over there. That migration was less troubled than the original one (it seemed to kill my server while it was happening). So the bottom line, in my opinion, was that somehow I had a poisoned domain that was ultimately cleansed by migrating it twice. Strange but true...