Multisite configuration

[Valerion]

local-host-names tell the server what domains are considered local, so if it does not exist here it will be bounced. Otherwise sendmail will do a MX lookup and forward the mail.

You need to ensure sendmail and Scalix routing works fine for inter-server email. Send a test message to +TEST on the remote server/ The Test Server will automatically bounce it with a reply, showing that routing works. Test this for all mailnodes. If routing doesn't work, you won't be able to dp anything else.

Once routing works, you can check dirsync to see if that works out at all. The best way to check is in the audit logs.

All inter-server emails, dirsyncs, etc are via port 25, so you just need to check for traffic there. Your sendmail logs can also tell you what happens.

Unfortunately there's so many variables it's very hard to debug this on a forum, you often need access to the machines in question to track down the issues. I would suggest you talk to a reseller.

[djtremors]

local-host-names tell the server what domains are considered local, so if it does not exist here it will be bounced. Otherwise sendmail will do a MX lookup and forward the mail.

I'm aware of that and know how sendmail works but if both sendmail servers assume they are domain.com locally, they won't pass any emails to any other server as sendmail isn't multi server aware, only scalix is aware that the same domain exists on other servers with other user accounts...

Code: Select all

[root@scalix1 mail]# omsend -u "scalix1 user" -p abc123 -s test -t "+TEST/mailnode"

Connected to scalix1

Logged on as scalix1 user /scalix1

Address Unknown  to  :  +TEST /mailnode


Err... did I do it wrong?
I can email from scalix1 to root@scalix2.domain.com but i can't do it as root@domain.com as it tries the MX record mail server which isn't scalix.

All inter-server emails, dirsyncs, etc are via port 25, so you just need to check for traffic there. Your sendmail logs can also tell you what happens.

I have emails every 15minutes betwen scalix@scalix1.domain.com -> scalix@scalix2.domain.com and vice verse on the other server. But I don't see syncs or any updates user lists on SAC
I saw a post elsewhere about uninstalling SAC+RES and installing RES only through the installer, but even in custom mode I don't see an uninstall for RES only that Scalix Message System (or something) which contains both and when trying to readd RES it prompts for a name of the ____ server which i put the scalix1 server (doing this on scalix2.. also not sure the name it said for the ____ part)..

So there is nobody who has done a walkthrough of 2 server setup with a centralise management server.. besides 1 server being typical install, a 2 server with dirsync and centralised administration would be next most common setup i'm sure. I would do a howto walkthrough if I could only get this to work as expected.

[Valerion]

local-host-names tell the server what domains are considered local, so if it does not exist here it will be bounced. Otherwise sendmail will do a MX lookup and forward the mail.

I'm aware of that and know how sendmail works but if both sendmail servers assume they are domain.com locally, they won't pass any emails to any other server as sendmail isn't multi server aware, only scalix is aware that the same domain exists on other servers with other user accounts...

Once dirsync works, Scalix will contain all the addresses on all the servers, and will have sendmail pass it onwards, so to sendmail it will look to be local on the server. Scalix will be responsible for passing it on.

Code: Select all

[root@scalix1 mail]# omsend -u "scalix1 user" -p abc123 -s test -t "+TEST/mailnode"

Connected to scalix1

Logged on as scalix1 user /scalix1

Address Unknown  to  :  +TEST /mailnode


Err... did I do it wrong?
I can email from scalix1 to root@scalix2.domain.com but i can't do it as root@domain.com as it tries the MX record mail server which isn't scalix.

Mmm ... the syntax works on my test server. Try sending to a user on the remote machine using the O/R address (User Name/mailnode) and if it arrives in the user's mailbox (use omlist and omread to check. omlogon and omlogoff may also be useful). Also send a reply email and see if that arrives.

All inter-server emails, dirsyncs, etc are via port 25, so you just need to check for traffic there. Your sendmail logs can also tell you what happens.

I have emails every 15minutes betwen scalix@scalix1.domain.com -> scalix@scalix2.domain.com and vice verse on the other server. But I don't see syncs or any updates user lists on SAC
I saw a post elsewhere about uninstalling SAC+RES and installing RES only through the installer, but even in custom mode I don't see an uninstall for RES only that Scalix Message System (or something) which contains both and when trying to readd RES it prompts for a name of the ____ server which i put the scalix1 server (doing this on scalix2.. also not sure the name it said for the ____ part)..

So there is nobody who has done a walkthrough of 2 server setup with a centralise management server.. besides 1 server being typical install, a 2 server with dirsync and centralised administration would be next most common setup i'm sure. I would do a howto walkthrough if I could only get this to work as expected.

For now you can ignore SAC and RES, they are not used for dirsyncs. Once you have dirsyncs working you can see if you can get the centralized management to work.

Add a -v to your agreement (ommodds) and set the relevant error level. This will let the Error Notification User (ENU) on both servers know what dirsync is doing. This is frequently helpful to see errors, etc.

[djtremors]

OK, this is strange.. after a week or so of stuffing around. I suddenly have my 2 users and 2 sxadmins listed on my main (scalix1) server... I don't know if adding domain.com to the local-host-names file and restarting sendmail then waiting for the 15min sync to happen made it work or not but it's suddenly there after such a long struggle. I had taken off the local-host-names settings out which is why I'm not sure if that did it... hmm...

I can't figure out how to email the other user using omsend.. both servers have the name mailnode and I don't know if thats a bad thing considering this name is a mailnode on both servers.... I go to webmail and i can't see the other name on the TO list...argh...frustrating.

if a "mailnode" is on each server, does it make a difference if they are labelled the same? I see people with example commands using mailnodeA and mailnodeB, is this purely reference or basically to distinct they are different nodes because mine are a mailnode on each server but it's shown to be the same name.

[Valerion]

You need different mailnodes names on each server. Otherwise Scalix has no way to tell users from one server apart from users on another server. VERY important.

[djtremors]

Aaaah.. see thats where learning from other posts is bad..
viewtopic.php?t=1839&highlight=directory+sync+omaddrt

How in the installation are you suppose to change the name of the mailnode, i think from memory it was automated and just says it's creating it.

Can i change the mailnode name via command line?

PS: thanks heaps for clearing things up, i will eventually document a multiserver setup one day soon

[djtremors]

Ok now i'm trying to figure out the true name of my mailnodes because in SAC they are the same name as the server.. ie server name=scalix1 mailnode=scalix1

but when i run omshowds -q all on scalix1 I get this

Code: Select all

UNIX       internet                                        MIME
UNIX       internet,tnef                                   TNEF
SMINTFC    mailnode2                                       scalix@scalix2.domain.com
LOCAL      scalix1
SMINTFC    scalix2,scalix2                                 scalix@domain.com


It displays "mailnode2".. is that the true name or is that what I should use on the omaddds commands etc..?? As you can see I used "scalix2" as the mailnode for the scalix2 dirsync but this is when I noticed "mailnode2" above it.

i'm lost.

[djtremors]

OK omsend now works with

Code: Select all

 omsend -u "scalix1 user" -p abc123 -s test -t "scalix2 user/mailnode2"


and I see logs of it passing through sendmail logs

Code: Select all

Feb 14 12:09:13 scalix2 sendmail[12948]: m1E19DFm012948: from=scalix@scalix2.domain.com, size=1646, class=0, nrcpts=1, msgid=H0000067000020d3.1202951353.scalix2.domain.com, proto=ESMTP, daemon=MTA, relay=localhost.localdomain [127.0.0.1]

Feb 14 12:09:14 scalix2 sendmail[12950]: m1E19DFm012948: to=scalix@scalix1.domain.com, delay=00:00:01, xdelay=00:00:01, mailer=esmtp, pri=121646, relay=172.16.200.3. [172.16.200.3], dsn=2.0.0, stat=Sent (Message accepted for delivery)

But nothing ends up in the mailbox of the user when checking in SWA. I thought I was almost there...
Oh also, i still see no audit logs of syncs on scalix1 when i see scalix2 requesting updates.... *sigh*

[Valerion]

All is not lost Remove your I would suggest re-installing one of the machines (at least) and respecifying the mailnode. It is asked for during the install, before the mailstore gets created, and the default is based on the hostname. Then set up the routing between the mailnodes with omaddrt, then test mail flow. Then do your dirsyncs.

[djtremors]

Would you suggest using "custom" install, as this prompted a little too much for me as it asked LDAP settings etc.. which i don't know if I selected one too many options to configure or not.

I might just do both again and note the settings to help others configuring this kind of setup.

[Valerion]

Look at the top screenshot on p.49 of the installation guide. There's where you specify the mailnode.

[djtremors]

On scalix2 server (remote), used custom and default entries except for below :

Code: Select all

=== Server Configuration ===
-> Enter the name of the primary mailnode on this server [scalix2]: mailnode2

=== RES Configuration ===
Please enter the fully qualified hostname of the machine where the Scalix Management Services is or will be instald.
-> Hostname [scalix2.domain.com]: scalix1.domain.com



Also used my original command to add routes and dirsyncs except used proper mailnode# numbers.

yet i still can't sync. I noticed now in this installation i now can't webmail on either server.. only the main server and only users on mailnode1. if I add users to mailnode2 for some reason they can't login... F%$^@

[Valerion]

It should work, if you do all the steps correctly. I think this is something different in your setup. There's often something small in an environment that's different that causes things to go wrong. I think you should get a Scalix reseller to do this for you. It's going to be hard to debug in the forum without seeing all the messages flowing, etc.

[djtremors]

well, actually all looks well and even though the audit logs appear to be only seen on the scalix2 server and nothing on scalix1 as from my understanding there should be, the only thing that doesn't seem to work is that SWA and SAC on scalix2 can't log in but if you look at SAC on scalix1 it shows all users and sxadmins which i think should be right except that SWA on both should work.

When I list the users on both servers they show they are homed correctly.

Code: Select all

[root@scalix1 scalix-11.3.0-GA]# omshowu -m all
sxadmin /mailnode1/CN=sxadmin
sxqueryadmin /mailnode1/CN=sxqueryadmin
test user1 /mailnode1/CN=test user1


Code: Select all

[root@scalix2 scalix-11.3.0-GA]# omshowu -m all
sxadmin /mailnode2/CN=sxadmin
sxqueryadmin /mailnode2/CN=sxqueryadmin
test user2 /mailnode2/CN=test user2


So really, if I can users authenticate on scalix2 for SWA then I'm pretty much sweet anyway.... just anything from mailnode2 cannot login to SAC2 or SWA2 (if you know what i mean)...
users from mailnode1 can login to SWA on scalix1 fine but just not users of mailnode2.

any ideas on this issue then?

[Valerion]

Once dirsync is sorted out SWA will do cross-server logins. It uses the address book records to determine the mailnode, and from there the server.