[Mailman-Users] Setting up a failover system for Mailman
The List Server Administrator at UNH
listadm at paradox.unh.edu
Wed Aug 18 17:23:56 CEST 1999
Folks,
We maintain a failover system for our primary web server system.
Using "rdist" we copy over the server's data tree every hour to
the failover system. Should the primary system fail, we can put
the failover system on-line and have it answer to that address.
The idea is that (we hope) at most only an hour's worth of
updates made by our content providers (faculty, staff, and
students) would be lost.
We'd like to setup something similar for our mailing list
services. I was wondering if anyone else has already gone down
this road and would appreciate any advice. One of the things I'm
concerned about is that a list server (unlike a web server?) is
very 'stateful'. That is, lists go through states of waiting to
have a message to be distributed, in process of having the
message delivered, and then filing away the message that was
delivered, archiving copies, etc. For example, I would expect
that Mailman, on start-up, is designed to detect if it was
interrupted, say by a system crash, in the delivery of a posting,
and would queue that posting for retry.
So my question is, would it be strictly necessary to stop the
server each time before capturing the data tree in order to be
sure the server's files would be in a stable enough state?
To look at it another way, this is really just a backup
procedure (although one that is potentially done much more
frequently then the usually nightly backups). So does your site
bother to shutdown Mailman before doing your nightly backups?
Is there the danger of having a restored server data tree that will
not run if this is not done?
TIA for any insights or ideas.
Cordially,
The List Server Admin
list.admin at unh.edu
(currently Bill Costa)
More information about the Mailman-Users
mailing list