Topic: https://mail/server-manager not working after upgrading to 9.0

[cyberwatcher]

when accessing the console via an ssh session it specifically says when accessing the url from a web browser us:
https://mail/server-manager
 

so RequestedDeletion appreciate your help and everyone else but you telling me that SME never worked that was is not correct.

let me know if you need anymore logs or anything. I am still trying to figure this out.

[guest22]

when accessing the console via an ssh session it specifically says when accessing the url from a web browser us:
https://mail/server-manager

This is a very confusing sentence to me.

Your domain is cyberwatch.com, nothing else. Forget about mail.whatever.  Let's focus on a normal functioning of the server and leave mai setup out of the mix here.

You granted access to your local sub networks to server manager. Did you enter the correct netmasks?

[cyberwatcher]

yes I did that. It came over from the restore but I checked everything over and over. I will say this, and I should have mentioned it and would have but I didnt remember until today, I was able to access the server-manager before doing the restore. Not after :/

cyberwatchers.com is working for everything else. mail flow working fine. it is weird. I even deleted my IE and mozilla folders from my computer so that I can access the server new with no cert. didnt work. I am going to connect my mac to the DMZ port the mail server is on (172 network) and see if directly connecting to the network it is on lets me in.

[cyberwatcher]

RESOLVED!!!

it worked when connecting directly to the subnet my mail server is on. So when looking at my firewall policy lol yep that was it. Its working now. I am not sure why but the upgrade uses port 80 when accessing the server-manager even thought I am using https.... I never needed that port to go from "trust to dmz" before. I saw port 80 being blocked each time I access the server via https://mail/server-manager.

Why is that? I am sure there is a logical reason. Thanks for you help and if I hit up on something interesting let me know.

so the fix was allow port 80 from my trust LAN out to my DMZ where my server is located.

do you close this case or can I? thanks again for the help.

[janet]

cyberwatcher, RequestedDeletion et al

[root@mail etc]# hostname -f
mail.cyberwatchers.com

root@mail etc]# db domains show
cyberwatchers.com=domain
    Content=Primary
    Description=Primary domain
    Nameservers=localhost
    Removable=no
    SystemPrimaryDomain=yes

The mail hostname name works/resolves as that is how the server is configured/named.
Instead of the servername being server123 it is named mail, so therefore http://mail/server-manager will work on the LAN,
as would http://server123/server-manager.

[janet]

cyberwatcher

I realise you now say this is resolved ie as a local firewall issue.

I will say this, and I should have mentioned it and would have but I didnt remember until today, I was able to access the server-manager before doing the restore. Not after :/

Did you remove any custom templates that were created for sme7 ?
You should have, & then recreated them for sme9.

Also there are significant changes between sme7 & sme9 & somewhere along the way I recall changes to how server manager was being accessed (local port or something....), so maybe some unexpected effect.

Referring to my earlier post, perhaps naming your server as "mail" is not the best idea, it can create confusion.

The URL mail.cyberwatchers.com will automatically be created by the system, refer server manager Configuration overview page.

[cyberwatcher]

no I didnt recreate any templates or remove any. I did this as a new install on another server. I still have the old one if incase I need to rebuild this one. I could always do that and this time name it mail1 or something. Its working fine now. I have added my OSSEC agent on it with is reporting fine to my OSSEC server. the only thing I need to do now is get syslog messages to pass to my syslog-ng server. I realize rsyslog is being used now version 9.0 and will be opening up a ticket.

for now it seems to be fine but now I am worried since it is named mail. I have had it that way for years on the other versions. renaming it would be more work as I would have to change the MX record at my name provider. not a big deal I guess.

what issues should I be looking for now since it is currently names mail?

[janet]

cyberwatcher

Your SystemName appears to be mail.
Just to be sure what is output of
config show SystemName

no I didnt recreate any templates or remove any. I did this as a new install on another server.

Yes but you said you did a restore of the sme7 backup to a fresh install of sme9, so any custom templates that were on sme7 will now be on sme9.
You should delete those after the restore, or recreate/check to see if the custom templates match the template code being used natively in sme9.

You can log in to the console (as admin) & run through the Configure this server screens & simply rename the server to something less "conflicting"
eg mymachine1 or whatever, anything other than mail, as that name conflicts (in a naming sense) with automatically created URL names.

You do not need your server to have a domain name of mail.yourdomain.com nor do you need external DNS to be configured for that, the sme server will respond to mail sent to yourdomain.com

I do not think it is causing serious issues, other than confusing those trying to help you.

[cyberwatcher]

ahhh ok I see. Yeah I expanded templates a few times for something I can't remember now. not many at all though. but I get it. so I will just change my nat rule on my firewall to point to the new name something other than mail.

If I do find templates that are 7.0 ones how can I get the 9.0 ones to replace ? thanks for the help and sorry for the confusion.

[janet]

cyberwatcher

If I do find templates that are 7.0 ones how can I get the 9.0 ones to replace ? thanks for the help and sorry for the confusion.

Theoretically any custom templates you find (in the custom templates tree) on your sme9 server, will be sme7 ones, unless you have been creating new ones on sme9.
So you could move any/all those custom templates to a safe place (folder) & then copy the original sme9 templates from the templates tree to the custom template tree & modify as necessary based on your requirements & what it is you were doing in the sme7 custom templates (use them as a guide).

Note I have seen many instances where the templates in older & newer versions of sme server are different & older templates do cause problems (errors etc), so it is best to recreate your custom templates based on the sme9 templates.
Some templates remain the same/similar, & some do not.

Follow with
signal-event post-upgrade
signal-event reboot

[cyberwatcher]

ok thanks I will research and figure that out. makes sense. this case can be closed though If I have any issues I will get with you guys. thanks for what you do.

[CharlieBrady]

That is a different URL then you posted as your main problem/title. But in both cases you should use a FQDN.

Or an IP address, which would eliminate the possibility of a name resolution problem.

OP should look in /var/log/httpd/error_log and /var/log/httpd/admin_error_log. And remain calm, be friendly and polite, and provide more clear information.

[cyberwatcher]

Janet I have searched /etc/e-smith/templates-custom and templates-user-custom and don't see anything in there actually. I don't think it copied anything over. I remember expanding a template I think for user console to create vacation time and such but that did not get installed with this new version. I will install it later.

[cyberwatcher]

Charlie Brady yes I did use the local IP address when testing thanks. the logs which I included in previous posts I did add. I didnt see anything in the logs stating port 80 was being blocked.

I still find it odd that I had to enable 80 from one network segment to the segment my mail server is on in order to get https to work. I didnt have to do that in last version. its working though so. thanks again.

[CharlieBrady]

Charlie Brady yes I did use the local IP address when testing thanks. the logs which I included in previous posts

You didn't include any logs or tcpdumps in any previous posts, as you can verify by just reading this thread. You didn't say anything about different network segments or anything about port 80.

We do not have a crystal ball and cannot read your mind.

I think it would be useful for you to read this essay:

http://www.chiark.greenend.org.uk/~sgtatham/bugs.html

I'm glad that your system is now working.