| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> RE: Failover with Oracle8 Parallel server
Patty,
I have experienced the problem you describe. What we did (you have to be on Unix for this), is to monitor the network card, when the script (or product like Openview) determines the card is down or not answering we have another network card answer for that ip address(I think ifalais is the command). When a request is passed to this new card, the sid you are trying to connect to will not be found (this is a much quicker error). You have to be careful when you do this because you could get duplicate IP's on the network if your NIC monitoring script falsely detects the other card being down and starts answering for that IP.
Michael Boligan
Sr. Database Admin
Healthvision
mboligan_at_healthvision.com
-----Original Message-----
From: Andrew Timpo [mailto:atimpo_at_mediaone.net]
Sent: Monday, September 18, 2000 10:55 PM
To: Multiple recipients of list ORACLE-L
Subject: Re: Failover with Oracle8 Parallel server
We experienced the same thing at our site when testing failover with OPS. Basically, you are running into a TCP timeout problem. I'm no networking god, but my understanding is that when you issue a packet to a host that's "dead" (not responding on the networking layer), there is a default timeout that TCP will wait to get a response before declaring that connection "down". I called support up about it & they claimed nothing could be done (not that I think otherwise; I just don't know enough about networking to validate or dispute their claim; the MPP group at support is great). Everyone I had spoken to about altering the default TCP timeout suggested that it would have bad impacts if the kernel were altered in this manner (even assuming it could be done, which I don't know).
Our normal tnsnames.ora part goes like this:
service.world=
(address_list=
(address=
(host=node_1)(port=1525)(sid=sid_name)(protocol=tcp)
(host=node_2)(port=1525)(sid=sid_name)(protocol=tcp)
...
I wrote a script such that if a node death was detected, a new copy of tnsnames.ora was copied in so that the living node would be listed first...
service.world=
(address_list=
(address=
(host=node_2)(port=1525)(sid=sid_name)(protocol=tcp)
(host=node_1)(port=1525)(sid=sid_name)(protocol=tcp)
...
This way, if node_1 was down, it would skip the TCP timeout issue, and connect to something that would hook up immediately.
Hope this helps!
Andrew Timpo
Penny Keefe wrote:
> Anyone have any experience with Oracle8 Parallel server? We are having
> difficultly failing over when the network connection is pulled from the
> server (similulating a network card or connection failure). To be more
> specific-- We have a program that uses OCI to connect to the database.
> When we pull the network connection it takes 3 minutes before sqlnet
> sends an error message. I was wondering if there is anyway to shorten
> the response time to a minute or less. The verison of Oracle is 8.0.5
> and we use local hostnaming (tnsnames) to connect to the server.
>
>
>
>> To REMOVE yourself from this mailing list, send an E-Mail message
> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> San Diego, California -- Public Internet access / Mailing Lists
> --------------------------------------------------------------------
-- Author: Andrew Timpo INET: atimpo_at_mediaone.net Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051 San Diego, California -- Public Internet access / Mailing Lists -------------------------------------------------------------------- To REMOVE yourself from this mailing list, send an E-Mail message to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You mayReceived on Tue Sep 19 2000 - 08:48:11 CDT
 |
 |