Loading ...
Sorry, an error occurred while loading the content.
 

High Availability

Expand Messages
  • Ramesh
    Hi All, Presently we have primary MX and backup MX servers, when primary goes down mails will be queued in secondary MX, once primary restored all messages
    Message 1 of 14 , Apr 6, 2014

      Hi All,

      Presently we have primary MX and backup MX servers, when primary goes down mails will be queued in secondary MX, once primary restored all messages pushed from backup MX to primary MX, messages are not lost. I would like to know any solution sending and receiving messages from backup MX when primary MX is down?

      Appreciate suggestion, recently due to major internet service down, we are not able to check mails or send mails.

      Thanks
      Ramesh 
    • Wietse Venema
      ... This is the wrong question on the wrong mailing list. The right question is I need a message store that is replicated in multiple locations . Once you
      Message 2 of 14 , Apr 6, 2014
        Ramesh:
        >Presently we have primary MX and backup MX servers, when primary
        >goes down mails will be queued in secondary MX, once primary restored
        >all messages pushed from backup MX to primary MX, messages are not
        >lost. I would like to know any solution sending and receiving
        >messages from backup MX when primary MX is down?

        This is the wrong question on the wrong mailing list.

        The right question is "I need a message store that is replicated
        in multiple locations". Once you have such a message store, any
        number of Postfix MTAs and mail clients can use it.

        Wietse
      • Ramesh
        Im sorry Wietse, Please let me know how to implement this, share me urls i will go through it how replication helps to solve our problem. Regards, Ramesh On
        Message 3 of 14 , Apr 6, 2014
          Im sorry Wietse,

          Please let me know how to implement this, share me urls i will go through it how replication helps to solve our problem.

          Regards,
          Ramesh
          On Monday, 7 April 2014 1:00 AM, Wietse Venema <wietse@...> wrote:
          Ramesh:

          >Presently we have primary MX and backup MX servers, when primary
          >goes down mails will be queued in secondary MX, once primary restored
          >all messages pushed from backup MX to primary MX, messages are not
          >lost. I would like to know any solution sending and receiving
          >messages from backup MX when primary MX is down?


          This is the wrong question on the wrong mailing list.

          The right question is "I need a message store that is replicated
          in multiple locations".  Once you have such a message store, any
          number of Postfix MTAs and mail clients can use it.

              Wietse



        • Wietse Venema
          ... To find solutions, open your favorite search engine and try cyrus mailbox replication , dovecot meailbox replication , and so on. Wietse A: Because it
          Message 4 of 14 , Apr 6, 2014
            Wietse:
            >The right question is "I need a message store that is replicated
            >in multiple locations". Once you have such a message store, any
            >number of Postfix MTAs and mail clients can use it.

            Ramesh:
            >Please let me know how to implement this, share me urls i will go
            >through it how replication helps to solve our problem.

            To find solutions, open your favorite search engine and try "cyrus
            mailbox replication", "dovecot meailbox replication", and so on.

            Wietse

            A: Because it messes up the order in which people normally read text.
            Q: Why is top-posting such a bad thing?
            A: Top-posting.
            Q: What is the most annoying thing in e-mail?
          • Miles Fidelman
            Hi Wietse, ... I ve been wondering about this too, and it strikes me that mailbox replication is only relevant to local delivery. What about replicating the
            Message 5 of 14 , Apr 6, 2014
              Hi Wietse,

              Wietse Venema wrote:
              > Wietse:
              >> The right question is "I need a message store that is replicated
              >> in multiple locations". Once you have such a message store, any
              >> number of Postfix MTAs and mail clients can use it.
              > Ramesh:
              >> Please let me know how to implement this, share me urls i will go
              >> through it how replication helps to solve our problem.
              > To find solutions, open your favorite search engine and try "cyrus
              > mailbox replication", "dovecot meailbox replication", and so on.
              >

              I've been wondering about this too, and it strikes me that "mailbox
              replication" is only relevant to local delivery. What about replicating
              the various intermediate mail queues? (My current HA setup is brute
              force - a failover virtual machine, with a completely replicated file
              system. But I've been looking for ways that are more granular, and that
              are easier to do across two separate data centers.)

              Thanks,

              Miles Fidelman




              --
              In theory, there is no difference between theory and practice.
              In practice, there is. .... Yogi Berra
            • Patrick Ben Koetter
              ... On Linux use DRBD to replicate mail queues between a pair of machines and crm to control a second Postfix instance that will be started locally to pickup
              Message 6 of 14 , Apr 6, 2014
                * Miles Fidelman <mfidelman@...>:
                > Hi Wietse,
                >
                > Wietse Venema wrote:
                > >Wietse:
                > >>The right question is "I need a message store that is replicated
                > >>in multiple locations". Once you have such a message store, any
                > >>number of Postfix MTAs and mail clients can use it.
                > >Ramesh:
                > >>Please let me know how to implement this, share me urls i will go
                > >>through it how replication helps to solve our problem.
                > >To find solutions, open your favorite search engine and try "cyrus
                > >mailbox replication", "dovecot meailbox replication", and so on.
                > >
                >
                > I've been wondering about this too, and it strikes me that "mailbox
                > replication" is only relevant to local delivery. What about
                > replicating the various intermediate mail queues? (My current HA
                > setup is brute force - a failover virtual machine, with a completely
                > replicated file system. But I've been looking for ways that are
                > more granular, and that are easier to do across two separate data
                > centers.)

                On Linux use DRBD to replicate mail queues between a pair of machines and crm
                to control a second Postfix instance that will be started locally to pickup
                any remaining mails once the partner machine dies.

                p@rick

                --
                [*] sys4 AG

                https://sys4.de, +49 (89) 30 90 46 64
                Franziskanerstraße 15, 81669 München

                Sitz der Gesellschaft: München, Amtsgericht München: HRB 199263
                Vorstand: Patrick Ben Koetter, Marc Schiffbauer
                Aufsichtsratsvorsitzender: Florian Kirstein
              • Nikolaos Milas
                ... Hmm, I think DRBD is only advised in cases where the net link between the replicated boxes is guaranteed and low-latency; so I guess probably this is not a
                Message 7 of 14 , Apr 7, 2014
                  On 7/4/2014 8:17 πμ, Patrick Ben Koetter wrote:

                  > On Linux use DRBD to replicate mail queues between a pair of machines and crm
                  > to control a second Postfix instance that will be started locally to pickup
                  > any remaining mails once the partner machine dies.

                  Hmm, I think DRBD is only advised in cases where the net link between
                  the replicated boxes is guaranteed and low-latency; so I guess probably
                  this is not a working solution between different data centers as
                  discussed here.

                  However, I can't suggest alternatives, I am afraid...

                  My instinct (though not always correct :-) ) tells me also that
                  near-real-time file sync (like using lsyncd with rsync) should not be a
                  suggested solution for queue replication.

                  One could investigate whether Apache Helix (http://helix.apache.org/)
                  can be a viable solution.

                  All the best,
                  Nick
                • Wietse Venema
                  ... Have you considered the following: - Inbound mail spends a fraction of a second in the queue. - Inbound mail spends days or weeks or more in the mailbox. -
                  Message 8 of 14 , Apr 7, 2014
                    Miles Fidelman:
                    > > To find solutions, open your favorite search engine and try "cyrus
                    > > mailbox replication", "dovecot meailbox replication", and so on.
                    >
                    > I've been wondering about this too, and it strikes me that "mailbox
                    > replication" is only relevant to local delivery. What about replicating
                    > the various intermediate mail queues? (My current HA setup is brute
                    > force - a failover virtual machine, with a completely replicated file
                    > system. But I've been looking for ways that are more granular, and that
                    > are easier to do across two separate data centers.)

                    Have you considered the following:

                    - Inbound mail spends a fraction of a second in the queue.

                    - Inbound mail spends days or weeks or more in the mailbox.

                    - If an MTA goes down, mail flows via alternate MX hosts.

                    - If the mailbox store goes down, then you have no mail.

                    That's why high availability focuses on the mailbox store,
                    not on the MTA in the middle.

                    Wietse
                  • Miles Fidelman
                    ... Well yes, in theory - but in practice we run a bunch of email lists, and I find that there are always cases where one or more destinations are temporarily
                    Message 9 of 14 , Apr 7, 2014
                      Wietse Venema wrote:
                      > Miles Fidelman:
                      >>> To find solutions, open your favorite search engine and try "cyrus
                      >>> mailbox replication", "dovecot meailbox replication", and so on.
                      >> I've been wondering about this too, and it strikes me that "mailbox
                      >> replication" is only relevant to local delivery. What about replicating
                      >> the various intermediate mail queues? (My current HA setup is brute
                      >> force - a failover virtual machine, with a completely replicated file
                      >> system. But I've been looking for ways that are more granular, and that
                      >> are easier to do across two separate data centers.)
                      > Have you considered the following:
                      >
                      > - Inbound mail spends a fraction of a second in the queue.
                      >
                      > - Inbound mail spends days or weeks or more in the mailbox.
                      >
                      > - If an MTA goes down, mail flows via alternate MX hosts.
                      >
                      > - If the mailbox store goes down, then you have no mail.
                      >
                      > That's why high availability focuses on the mailbox store,
                      > not on the MTA in the middle.
                      >
                      >

                      Well yes, in theory - but in practice we run a bunch of email lists, and
                      I find that there are always cases where one or more destinations are
                      temporarily unavailable - so there are various messages that will hang
                      around for a while. So HA for the queues is not unreasonable to think
                      about.

                      Miles

                      --
                      In theory, there is no difference between theory and practice.
                      In practice, there is. .... Yogi Berra
                    • Wietse Venema
                      ... You change the topic of the discussion and then claim some contradiction. Wietse
                      Message 10 of 14 , Apr 7, 2014
                        Miles Fidelman:
                        > > Have you considered the following:
                        > >
                        > > - Inbound mail spends a fraction of a second in the queue.
                        > >
                        > > - Inbound mail spends days or weeks or more in the mailbox.
                        > >
                        > > - If an MTA goes down, mail flows via alternate MX hosts.
                        > >
                        > > - If the mailbox store goes down, then you have no mail.
                        > >
                        > > That's why high availability focuses on the mailbox store,
                        > > not on the MTA in the middle.
                        >
                        > [talking about OUTBOUND mail which was not the subject of this thread]

                        You change the topic of the discussion and then claim some
                        contradiction.

                        Wietse
                      • Miles Fidelman
                        ... Not to be argumentative or anything, but... original query was: Presently we have primary MX and backup MX servers, when primary goes down mails will be
                        Message 11 of 14 , Apr 7, 2014
                          Wietse Venema wrote:
                          > Miles Fidelman:
                          >>> Have you considered the following:
                          >>>
                          >>> - Inbound mail spends a fraction of a second in the queue.
                          >>>
                          >>> - Inbound mail spends days or weeks or more in the mailbox.
                          >>>
                          >>> - If an MTA goes down, mail flows via alternate MX hosts.
                          >>>
                          >>> - If the mailbox store goes down, then you have no mail.
                          >>>
                          >>> That's why high availability focuses on the mailbox store,
                          >>> not on the MTA in the middle.
                          >> [talking about OUTBOUND mail which was not the subject of this thread]
                          > You change the topic of the discussion and then claim some
                          > contradiction.
                          Not to be argumentative or anything, but... original query was:
                          "Presently we have primary MX and backup MX servers, when primary goes
                          down mails will be queued in secondary MX, once primary restored all
                          messages pushed from backup MX to primary MX, messages are not lost. I
                          would like to know any solution sending and receiving messages from
                          backup MX when primary MX is down?"

                          Which sure looks like it includes outbound ("sending") as part of the topic.

                          Miles
                        • Wietse Venema
                          ... He describes the flow of email for domains that have MX records with the names of his MX hosts (when the primary MX is down, mail queues on the secondary
                          Message 12 of 14 , Apr 7, 2014
                            Miles Fidelman:
                            > Wietse Venema wrote:
                            > > Miles Fidelman:
                            > >>> Have you considered the following:
                            > >>>
                            > >>> - Inbound mail spends a fraction of a second in the queue.
                            > >>>
                            > >>> - Inbound mail spends days or weeks or more in the mailbox.
                            > >>>
                            > >>> - If an MTA goes down, mail flows via alternate MX hosts.
                            > >>>
                            > >>> - If the mailbox store goes down, then you have no mail.
                            > >>>
                            > >>> That's why high availability focuses on the mailbox store,
                            > >>> not on the MTA in the middle.
                            > >> [talking about OUTBOUND mail which was not the subject of this thread]
                            > > You change the topic of the discussion and then claim some
                            > > contradiction.
                            > Not to be argumentative or anything, but... original query was:
                            > "Presently we have primary MX and backup MX servers, when primary goes
                            > down mails will be queued in secondary MX, once primary restored all
                            > messages pushed from backup MX to primary MX, messages are not lost. I

                            He describes the flow of email for domains that have MX records
                            with the names of his MX hosts (when the primary MX is down, mail
                            queues on the secondary MX, from which it's sent to the primary).
                            In other words, he describes inbound email.

                            Outbound mail. on the other hand, is sent to the remote MX hosts
                            of remote destination domains. Those remote MX hosts are not the
                            MX hosts that he is talking about.

                            Wietse
                          • tejas sarade
                            I don t see any reason to complicate things by implementing HA solution, when you can simply have multiple MX records. ... down mails will be queued in
                            Message 13 of 14 , Apr 7, 2014
                              I don't see any reason to complicate things by implementing HA solution, when you can simply have multiple MX records.


                              On Mon, Apr 7, 2014 at 12:35 AM, Ramesh <itsramesh_s@...> wrote:
                              >
                              >
                              > Hi All,
                              >
                              > Presently we have primary MX and backup MX servers, when primary goes down mails will be queued in secondary MX, once primary restored all messages pushed from backup MX to primary MX, messages are not lost. I would like to know any solution sending and receiving messages from backup MX when primary MX is down?
                              >
                              > Appreciate suggestion, recently due to major internet service down, we are not able to check mails or send mails.
                              >
                              > Thanks
                              > Ramesh
                            • Jose Ildefonso Camargo Tolosa
                              ... Hi! ... As Wietse already said, you can just have a replicated message store, as long as you accept that outgoing mail queue (most sites have some messages
                              Message 14 of 14 , Apr 8, 2014



                                On Sun, Apr 6, 2014 at 2:05 PM, Ramesh <itsramesh_s@...> wrote:

                                Hi All,


                                Hi!
                                 
                                Presently we have primary MX and backup MX servers, when primary goes down mails will be queued in secondary MX, once primary restored all messages pushed from backup MX to primary MX, messages are not lost. I would like to know any solution sending and receiving messages from backup MX when primary MX is down?

                                Appreciate suggestion, recently due to major internet service down, we are not able to check mails or send mails.

                                As Wietse already said, you can just have a replicated message store, as long as you accept that outgoing mail queue (most sites have some messages lying there, waiting to retry) and *maybe* one or two messages in the intermediate queues (highly unlikely) will be "unavailable" until you restore primary, and could be potentially lost if primary dies.

                                Otherwise you would need to replicate queue directories, likely using DRBD.  Yes, you can use DRBD over long-distance links, but you will have increased latency and reduced write performance (search for DRBD Proxy for an explanation).  I have implemented DRBD using softlayer's private network, but only for systems where reads/writes ratio is high.

                                Ildefonso.

                              Your message has been successfully submitted and would be delivered to recipients shortly.