Scalix 11.1 CE upgraded -- Webmail login fails

[rgmhtt]

Here is my situation:

I rsynced my /var/opt/scalix from my Scalix 10 live server (after stopping all Scalix processes) to my new Centos 5 server. One thing that MIGHT be a problem, is I did not created the scalix user first, and everything rsynced was for userid 100, group 101. Even after I created the user, ls -ls was still reporting the numbers, not the name, scalix.

I ran ./scalix-installer --override=rhel5

All testing after install was done with the router disconnected from my ISP. So the scalix server had a network but no internet. This box is also my DNS server, so it had all it needed.

It seems to have converted my message store. I can get into sac and see the new version number and new 11.1 stuff.

I go to webmail and the login (with the same userid/password as used in sac) fails. Bad userid or password (note passwords are case sensitive. Try again.). I try a userid that does not have sac auth and get the dreaded "The evaluation period for this time-limited product has expired". I restart scalix-tomcat as advised in the FAQ on the wiki, and then I get an http 503 error.

At first I was getting SELinux errors. I tried the realign and reboot, but still errors; I was running SELinux in permissive mode. So I disabled SELinux. Still webmail would not work.

So now what? Is this the Tomcat error that Valerion mentioned yesterday in: viewtopic.php?t=8340 ?

So if I do the uninstall as recommended in that reply, Do I keep the message store and user database? Note that my Scalix 10 server is back up and getting and sending mail. So I will have to move the data over again before attempting an install again. Can I do that? Will I get the conversion?

Help please.

[les]

Here is my situation:

I rsynced my /var/opt/scalix from my Scalix 10 live server (after stopping all Scalix processes) to my new Centos 5 server. One thing that MIGHT be a problem, is I did not created the scalix user first, and everything rsynced was for userid 100, group 101. Even after I created the user, ls -ls was still reporting the numbers, not the name, scalix.

I ran ./scalix-installer --override=rhel5

All testing after install was done with the router disconnected from my ISP. So the scalix server had a network but no internet. This box is also my DNS server, so it had all it needed.

It seems to have converted my message store. I can get into sac and see the new version number and new 11.1 stuff.

I go to webmail and the login (with the same userid/password as used in sac) fails. Bad userid or password (note passwords are case sensitive. Try again.). I try a userid that does not have sac auth and get the dreaded "The evaluation period for this time-limited product has expired". I restart scalix-tomcat as advised in the FAQ on the wiki, and then I get an http 503 error.

At first I was getting SELinux errors. I tried the realign and reboot, but still errors; I was running SELinux in permissive mode. So I disabled SELinux. Still webmail would not work.

So now what? Is this the Tomcat error that Valerion mentioned yesterday in: viewtopic.php?t=8340 ?

So if I do the uninstall as recommended in that reply, Do I keep the message store and user database? Note that my Scalix 10 server is back up and getting and sending mail. So I will have to move the data over again before attempting an install again. Can I do that? Will I get the conversion?

Help please.

I cant find an old post i did to answer this question.....

but in this scenario you should do it like this.....

assuming you are going from centos 4 to 5 or similar.

1. Upgrade to 11 first on your existing server. It should be fairly painless.
2. Install a default scalix 11 installation on the new server.
3. Make sure you set all the options up identically as you did before, especially mailnode.
4. Stop scalix on both.
5. migrate directories and configs across....
/var/opt/scalix/xx
/etc/opt/scalix-tomcat
/etc/opt/scalix
If you setup the configs identically then you probably only need /var/opt/scalix/xx and just check the rest.
6. Start scalix on the new server.

p.s. before migrating to the new server you may also want to backup the "virgin" /var/opt/scalix directory as a fallback in case something goes wrong.

If you do it this way you wont get issues with uid or gid and its a smoother transition.

This is how i did my centos4/scalix 10 to centos 5/scalix 11 upgrade to new hardware.

Hope that helps.

[rgmhtt]

I cant find an old post i did to answer this question.....

but in this scenario you should do it like this.....

assuming you are going from centos 4 to 5 or similar.

1. Upgrade to 11 first on your existing server. It should be fairly painless.

Sorry, but I lack confidence in things to even consider this. Too many things might be old/out-of-sync, and the upgrades fail and no way back to a working server.

I have seen it, lived it, all to many times. Granted this will be my first Scalix upgrade, but...

I can probably get back to a 'clean' system easily: Uninstall all of Scalix 11. rm everything in /var/opt/scalix. Create any userids I need besides scalix. rsync ALL necessary directories (you mention /etc/opt/ scalix and scalix-tomcat are these used in Scalix 10?)

Then I can try again.

2. Install a default scalix 11 installation on the new server.
3. Make sure you set all the options up identically as you did before, especially mailnode.
4. Stop scalix on both.
5. migrate directories and configs across....
/var/opt/scalix/xx
/etc/opt/scalix-tomcat
/etc/opt/scalix
If you setup the configs identically then you probably only need /var/opt/scalix/xx and just check the rest.
6. Start scalix on the new server.

Sounds reasonable but based on a iffy premise.

But it looks like:

viewtopic.php?t=8464

Is having much the same problems as me.

[les]

I cant find an old post i did to answer this question.....

but in this scenario you should do it like this.....

assuming you are going from centos 4 to 5 or similar.

1. Upgrade to 11 first on your existing server. It should be fairly painless.

Sorry, but I lack confidence in things to even consider this. Too many things might be old/out-of-sync, and the upgrades fail and no way back to a working server.

I have seen it, lived it, all to many times. Granted this will be my first Scalix upgrade, but...

I can probably get back to a 'clean' system easily: Uninstall all of Scalix 11. rm everything in /var/opt/scalix. Create any userids I need besides scalix. rsync ALL necessary directories (you mention /etc/opt/ scalix and scalix-tomcat are these used in Scalix 10?)

Then I can try again.

2. Install a default scalix 11 installation on the new server.
3. Make sure you set all the options up identically as you did before, especially mailnode.
4. Stop scalix on both.
5. migrate directories and configs across....
/var/opt/scalix/xx
/etc/opt/scalix-tomcat
/etc/opt/scalix
If you setup the configs identically then you probably only need /var/opt/scalix/xx and just check the rest.
6. Start scalix on the new server.

Sounds reasonable but based on a iffy premise.

But it looks like:

viewtopic.php?t=8464

Is having much the same problems as me.

those problems in that post are probably more to do with public folder permissions and the fact that the bb server is aborted than anything else.

When imap logs in, remembering that swa and imap use the same pam auth module whereas pop3 and mapi use a different one, it wants to list folders. If that list of folders fails it falls back to a standard "failed password" error, which is not the case. You see this when you enable full debuging of imap and watch the log as you attempt to login via imap.

centos 5 has no scalix 10 package.
scalix 10 to scalix 11 is a major upgrade and changes happen to the mail store, which yes are irreversible.
having said that if you backup properly beforehand, like all administrators should ;), then you could go back to version 10 if necessary.

I've done about 8 scalix upgrades from 10 to 11 without drama by just running the scalix installer and upgrading.

I've also done 2 or 3 migrations as indicated above without drama.

the scalix installer is first class. it will do everything for you, no need for you to go to the point of manually creating userids etc.
There might be one or 2 little things that need tweaking, such as tomcat virtual host files, but if you upgrade to 11.1.0 that shouldn't be necessary anymore. What you will need to change is how you setup the filtering via spamassassin which becomes easier in version 11, but thats documented in the wiki.

and yes /etc/opt/scalix and /etc/opt/scalix-tomcat do get used in 10.

That's my experience, its worked successfully for me on numerous occasions.

[rgmhtt]

:D


I unsinstalled Scalix 11.
Removed all the data structures.
Set up the scalix user id PROPERLY (including the Group scalix, that I got wrong last time).
rsynced all the scalix directories.

installed (really an upgraded) of Scalix 11.

Now everything is working.

Problem appears to be permissions on the files I rsynced. All is well now.

Well I still have a lot of tuning to do, and spamassasin...