Newby questions

[adhodgson]

Hi,

I have a couple of questions after trying out Scalix server for a couple of days:

1. What is the best way to configure Sendmail so that it doesn't do domain masquerading for local submissions? At current, I am getting messages re-written user@host.domain.com, whereas it should be user@domain.com. I believe it is Sendmail doing this, but have changed the DM option in sendmail to:

"DM" (I.E, null), and it still does it.

2. What are people using to sync two servers together? i am talking about doing a hot swap server for DR environments, the other server would only be used in the event of failure of the box. At current we are using a simple Rsync script.

3. What is the best way to create a PDL which is accessible publicly, e.g, sales? Also, we have some instances where I have separate mailboxes for some role accounts, so that duplication of messages/responses do not occur. What is the best way to set these up? We also need a way of generating an everyone list, is it better to set this up manually, so we don't end up adding role accounts etc to this? have seen the example script in the examples directory on the server.

Thanks.
Andrew.

[jch]

1. What is the best way to configure Sendmail so that it doesn't do domain masquerading for local submissions?

The easiest way to do domain masquerading is to edit /etc/mail/sendmail.mc and uncomment and edit the options there. After you've run "make" to rebuild sendmail.cf you'll also need to run omsendin to re-do the scalix config. Bear in mind, though, that masquerading isn't as important as it would be without Scalix since you tend to configure users with the address you want rather than a machine name, eg "jch@example.com" as opposed to "jch@scalix-server.example.com".

2. What are people using to sync two servers together? i am talking about doing a hot swap server for DR environments, the other server would only be used in the event of failure of the box. At current we are using a simple Rsync script.

You can use rsync to create a backup copy of a message store and use that if the worst comes to the worst, but if you're trying to protect against something like disk failure then use a RAID array. The enterprise version of Scalix has features that support high-availability using, for example, Heartbeat or Red Hat Cluster Services. In the fullness of time there'll be some documentation describing how to set that up, but in the meantime you can always ask your normal Scalix contacts.

3. What is the best way to create a PDL which is accessible publicly, e.g, sales? Also, we have some instances where I have separate mailboxes for some role accounts, so that duplication of messages/responses do not occur. What is the best way to set these up? We also need a way of generating an everyone list, is it better to set this up manually, so we don't end up adding role accounts etc to this? have seen the example script in the examples directory on the server.

PDLs are normally publically accessible and all you need to do is send a message to the PDL's internet address. Some of the role addresses at Scalix are done this way. The support mailbox here, on the other hand, is done as a separate account and at least one of the support guys uses Thunderbird so that he can have his home, work and support accounts all at the same time. Our "everyone" account is a very small PDL: it's three PDLs: everyone in the US, everyone in the UK and everyone in Germany. "Hierarchical" PDLs like this seem to be the easiest to manage. I have a plan for a differrent sort of PDL, one that's based on search criteria, but don't hold your breath for it, there's lots of other exciting things nearer the top of the pile.

jch

[adhodgson]

[quote="jch"][quote]1. What is the best way to configure Sendmail so that it doesn't do domain masquerading for local submissions?[/quote]
The easiest way to do domain masquerading is to edit /etc/mail/sendmail.mc and uncomment and edit the options there. After you've run "make" to rebuild sendmail.cf you'll also need to run omsendin to re-do the scalix config. Bear in mind, though, that masquerading isn't as important as it would be without Scalix since you tend to configure users with the address you want rather than a machine name, eg "jch@example.com" as opposed to "jch@scalix-server.example.com".

At present the issue is that we are having the system generate the from: header as user@machine.domain.com, even though in Scalix it is set as user@domain.com - i think this is due to domain masquerading, but want to just switch it off.

[quote]2. What are people using to sync two servers together? i am talking about doing a hot swap server for DR environments, the other server would only be used in the event of failure of the box. At current we are using a simple Rsync script. [/quote]
You can use rsync to create a backup copy of a message store and use that if the worst comes to the worst, but if you're trying to protect against something like disk failure then use a RAID array. The enterprise version of Scalix has features that support high-availability using, for example, Heartbeat or Red Hat Cluster Services. In the fullness of time there'll be some documentation describing how to set that up, but in the meantime you can always ask your normal Scalix contacts.

Its not really HA we are after - just a off-site backup we can use if the system goes down. It would only be account data we want to copy, and any data relating to PDLs etc.

Thanks.
Andrew.

[jch]

For the masquerading, check /etc/sendmail.mc -- there are good comments in there about what you can set up and it should be obvious what needs to be done to turn it off again. Re-generate the sendmail.cf and re-run omsendin,

LVM snapshots and rsync (especially with --link-dest) are a good way of doing poor-man's remote mirroring. Run omsuspend, wait a few seconds for outstanding operations to complete, take the snapshot, kill the suspend (see man page for details) and then rsync the snapshot. If you take a snapshot of ~scalix you'll have most of what you need: SWA, SAC and sendmail configuration are the obvious ones you don't have. You can pay more money, serious money, and get long-distance, delayed mirroring from your disk array. I know of one large organisation that used to have a remote mirror with a two hour delay (two copies of a corrupt database aren't terribly useful).

jch

[adhodgson]

Hi,

Was due to /etc/hosts having the domain name set to localhost - sendmail thought it was local - removing this fixed it.

We would probably do a nightly backup using LVM (after looking at this today).

Thanks.
Andrew.