Find the answer to your Linux question:
Results 1 to 6 of 6
Hello, I am using exim mail server on RHEL4 linux system.Yesterday mail server got stuckup with some problem and all the mails got blocked, untill i restart the mail server.After ...
Enjoy an ad free experience by logging in. Not a member yet? Register.
  1. #1
    Just Joined!
    Join Date
    Aug 2007
    Posts
    5

    Thumbs up linux mail server failover


    Hello,
    I am using exim mail server on RHEL4 linux system.Yesterday mail server got stuckup with some problem and all the mails got blocked, untill i restart the mail server.After facing this problem,i am searching on setting failover plan for mail server and got some idea on fall-back mail server (Linux as a Backup E-mail Server) using MX record.
    Here,the backup server only cache the mails and store in backup server.But i need to set a backup(slave) mail server which should come up and serve as master at the time of failure of master mail server.

    Any idea will be greatly appreciated....Thanks in advance...

  2. #2
    Linux Engineer jledhead's Avatar
    Join Date
    Oct 2004
    Location
    North Carolina
    Posts
    1,077
    I guess that depends what you want.

    If you only want incoming mail to failover, then a backup mx would be fine. if the primary goes down the outside world will automagically try the backup.

    If you want to mirror your mailserver to have a backup (all user data and everything) then thats a little more difficult.

    you could use something like lvs and ldirectord or keepalived to handle the failover mechanism. this won't take care of syncing your mail server from one server to the other.

    post back which you want and maybe I can be more descriptive.

  3. #3
    Just Joined!
    Join Date
    Aug 2007
    Posts
    5

    linux mail server failover

    Thanks for your reply, jledhead. I want to set a backup mail server (to send and receive) which can be used as mail server at the time of real mail server failure...
    i don't want to sync the data automatically between two mail servers i.e i don't want any backup of my mail account.I just want to run a secondary mail server(to avoid disturbance in mai process) at the time of primary server failure.
    Please advise me..

  4. #4
    Linux Engineer jledhead's Avatar
    Join Date
    Oct 2004
    Location
    North Carolina
    Posts
    1,077
    if your intention is for your clients to not receive an interuption, then using lvs will be the way to go. but even if you users are smtp/pop, there will be times when mail comes in that isn't downloaded, so if one server crashes it takes that mail with it, thats why I mentioned replication. lvs in itself is a pretty complicated piece of software.

    you can use ipvsadm alone or use something like ldirectord to control it (this is what I do). and for the ip failover you can use heartbeat or keepalived. both are similar in setup. this site has lots of info and examples Ultra Monkey: Configuration:

    lvs is documented with lots of information LVS Documentation
    if you use it and need help, might be a good idea to read thru the howto and then sign up for the list so you can ask and get help.

  5. #5
    Just Joined!
    Join Date
    Aug 2007
    Posts
    5

    Arrow

    1) what are the basic constraints for LVS?
    2) shall i make clustering between two remote servers? i.e two public ip's

  6. #6
    Linux Engineer jledhead's Avatar
    Join Date
    Oct 2004
    Location
    North Carolina
    Posts
    1,077
    Quote Originally Posted by linuxlearner_ephron View Post
    1) what are the basic constraints for LVS?
    not sure what you mean. do you mean how will it work? lvs can cluster more then one node behind a single ip address (vip) so clients would use the vip to connect instead of going directly to the server. lvs would then decide where to send the client, based on routing rules and available servers.

    2) shall i make clustering between two remote servers? i.e two public ip's
    I don't see the benefit of clustering 2 public ip's. the point of clustering is to make more then appear as one, so using 2 public ip's would not be clustered.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •