Frederic -
Your plan sounds interesting and good luck to you, but I doubt whether you can
get any certain answers to your questions here. You and the customer should be
planning on extensive testing for these answers. You are looking for guarantees
and I doubt there are any.
No one in NetView knows anything about how the NM/Expert agent works, except
what you have told us. They would be the ones to ask about whether it can still
function with ovwdb and ovtopmd down. While the XOM/XMP APIs rely on pmd, so
they should still work, what does NM/Expert do once it gets the trap? Read the
databases? If so, then you may still have a problem and you need to consult the
people who built this application, not NetView. Even after that, you will
probably have to test what they tell you. You should be able to do that even
before you build your backup system.
As for your HA set up, it seems reasonable enough, but until you actually test
it, you will not know for certain it will work.
And as for support of your procedure, that is rather murky. You will not find
any such procedure in NetView documentation. Yes, people do transfer databases
and run reset_ci all the time without problems, but the success of that depends
on the databases being properly shutdown and backed up, the ftp (or whatever
transfer mechanism you use) working correctly, and the restore as well. So it is
highly unlikely, given all the places that you might make an error, that you
will get any guarantees from us. Will we take an APAR if you have a problem
doing this? Probably not.
If you don't like that answer then I think you already know how to open a
request to development .
James Shanks
Tivoli (NetView for UNIX) L3 Support
Frederic Mottiat <frederic_mottiat@BE.IBM.COM> on 08/30/99 11:40:48 AM
Please respond to Discussion of IBM NetView and POLYCENTER Manager on NetView
<NV-L@UCSBVM.UCSB.EDU>
To: NV-L@UCSBVM.UCSB.EDU
cc: (bcc: James Shanks/Tivoli Systems)
Subject: About pmd usage...
Hello,
A customer of us has NetView V5 running on an AIX box. On the same system,
the customer is using an application called NM/Expert. As he told us,
NM/Expert register itself to the NetView pmd daemon so it can use Netview's
facilities to issue SNMP gets and sets as well as receiving traps. This
avoid double polling activities on the net, once from Netview and the
second time from NM/Expert.
The customer told us he is using the XOM/XMP APIs to talk to pmd.
We would like to achieve a High Availability scenario, using QualixHA+. We
would setup a second NetView server, also with NM/Expert.
In order to enable a very short startup of the backup system, we would like
to take a snapshot of the NetView databases regularly and move them to the
second server.
On this second server :
- all interfaces are shutdowned (ipconfig down)
- all netview daemons are running except netmon, ovwdb, ovtopmd and
snmpCollect (this prevents netmon from trying to poll anything, prevents
updates in the topo and object DB and prevents collections from being run)
When we take a databases backup on the primary NetView, in order not to
backup corrupted DBs we will first stop netmon, ovwdb, ovtopmd and
snmpCollect, and then copy the DBs over to the secondary system. Fast
backups could be implemented, but reset_ci and ovtopofix will have to be
ran on the secondary server after the copy is done. Does it seems
reasonnable ?
First question : while netmon, ovwdb, ovtopmd and snmpCollect are stopped
for the copy process, will NM/Expert still be able to ask pmd, via the
XOM/XMP APIs, to execute snmp queries ? Will it still be able to receive
traps as well ?
The second server as identical IP addresses and hostname as the primary
server. In case of failover, we activate the interfaces again, start back
netmon, ovwdb, ovtopmd and snmpCollect, working thus on the copied
databases. This give us a "warm" start of NetView, as starting those 4
daemons should be done in a few seconds only.
Second question : suppose our copied databases on the secondary server is 1
week old, and the node down interval is set to 7 days. Let us supposed that
there was 1 node marked as down in the topo DB just before the copy of the
DB is taken. What happens when netmon, ovwdb and ovtopmd restarts ? Does it
consider the node as being down for more than 7 days, and thus will it
delete it or will it poll it (status, etc) before ?
Last question : is the copy of the object and topo DBs a supported
operation (if we copy them to a NetView server with the same IP adresses or
at least in the same subnet) ?
Thank you in advance,
Frederic Mottiat
|