nv-l
[Top] [All Lists]

Re: Redundant/DR Netview Servers

To: nv-l@lists.tivoli.com
Subject: Re: Redundant/DR Netview Servers
From: Ray Schafer <schafer@tkg.com>
Date: Thu, 29 Jul 1999 01:16:04 -0400
Michael,

Yes this can be done.  It depends on how much in sync you want to keep these
servers.  It can get pretty complicated to keep them exactly in sync,  and may
be nearly impossible, but you can keep them darn close.

To begin with, you need to make sure that all of the nodes available in the
database of the primary are resolvable from the backup.   They must all be
reachable (pingable) from both servers.  SNMP connectivity must be the same as
well.

Status polling can be coordinated.  A good way to handle this is to use Mid
Level Managers.  All status and discovery can be off-loaded to them, and they
can forward their events to both Netview servers.  Agents can be configured to
forward their traps to the MLM's, which will in turn send these traps to both
servers via TCP instead of UDP so there is guaranteed delivery.   One netview
server can be configured to control the MLM's (status polling intervals, etc)
and the other server can be set up to poll only once per day - getting all of
it's status polls from the MLM(s).

I have successfully used rsync to keep /usr/OV/databases in sync, but it gets
tricky.  (rsync is a great utility written down under, I believe!!) Before the
rsync is started, any daemon (or customized program for that matter) that
causes updates to the database(s) to occur must be stopped.  The Netview
daemons netmon, ovtopmd, snmpCollect I believe are the only daemons you need to
deal with.  The read-write map must come down, and it's children  ipmap,
collmap, etc.   Alos don't forget about any rulesets that will be changing
database fields, etc.

On the receiving end of the rsync (the backup netview server), *ALL* daemons
must be stopped, and all maps must come down since the database under them is
being replaced.

Feel free to send me mail off-line for more details if you need it.


Michael Long wrote:

> We are looking at options for providing DR capabilities for a Netview
> customer.
>
> Currently they have 2 E250 Sun machines, with Netview 5.1.1 and want to use
> to act as a backup for the primary.
>
> Is anyone out there doing this:
> * is the second server polling?
> * What are the configuration requirements?
> * How do you update the second server with maps etc from the primary
> * Is it really stable
>
> Any other info you have on this would be appreciated.
>
> Thanks in advance.
> Michael Long
> Network Management Consultants Ltd
> PO. Box 38-161, Howick
> Auckland
> New Zealand
> Phone (6421) 610 480
> Fax      (6421)699 633

--
Ray Schafer                   | schafer@tkg.com
The Kernel Group              | Distributed Systems Management
http://www.tkg.com


<Prev in Thread] Current Thread [Next in Thread>

Archive operated by Skills 1st Ltd

See also: The NetView Web