Home › Forums › OS X Server and Client Discussion › Questions and Answers › Trying to get IP Failover set up in Tiger 10.4.2
- This topic has 6 replies, 2 voices, and was last updated 19 years, 8 months ago by
nicki.
-
AuthorPosts
-
August 5, 2005 at 3:26 pm #362688
nicki
ParticipantHi- I’m following some docs that I believe are incorrect from Mac about High Availability Administration. I’m trying to set up ip failover. IT gives directions on adding variables to hostconfig file basically, but most combinations of the variables seem to fail so I’m wondering if anyone has got this to work…. Here are the details:
Primary server: main Ip Address: 10.1.1.5, firewire IP= 192.168.1.10
Failover server: main IP: 10.1.1.6 Firewire: 192.168.1.11On primary server:
Edit hostconfig, add variables:
FAILOVER_BCAST_IPS=”192.168.1.11 10.1.1.6″On failover server:
FAILOVER_PEER_IP_PAIRS=”fw0:192.168.1.10″
FAILOVER_PEER_IP=”10.1.1.5″Which combination SHOULD work. The heart beat seems to always be looking on the primary ethernet address and NOT on the fw. The heart beat is fine until I unplug the ethernet on the main machine. When all machines are plugged into ethernet, the DNS lookup works fine. The only thing that does not work is the failover. And the failover log says “Unable to send to other machine: No route to host”
Any ideas?
August 5, 2005 at 4:38 pm #362690Anonymous
GuestMy SE and I have been trying to get htis working as well. One of the things that we discovered was wrong in the High Availability docs. They talk about a ProcessFailover script called Test at /Library/Failover/IP_Address/Test actually needs to be in /Library/IPFailover/IP_Address/Test. According to man ProcessFailover.
Also, it seems that the Script does need to exist and it does need to return True.
We are still working on getting all the kinks out of this on our server, but I thought this might be helpful information. I believe that it is correct. Of course YMMV. Standard disclosures, etc
javascript:emoticon(‘
‘)
August 5, 2005 at 4:47 pm #362691nicki
ParticipantHmm, I don’t even have that directory in /Library but I also didn’t see that in the docs. I think everything is working as far as failoverd works b/c it’s writing to the log when it can’t see the other server. I just can’t seem to force it to go over FW rather than ethernet to find the other server. I’m in a test network, too, with no firewall. I’ve heard that those docs are incorrect though. Would love some replacement docs
August 5, 2005 at 6:34 pm #362697Gary Bernstein
ParticipantI don’t remember if we had to create the /Library/IPFailover directory or not. I know that we had to create the /Library/IPFailover/YOURIPADDRESS/ directory and the Test file.
The section on this starts on page 14 or the High Availability docs.
HTH
-GaryAugust 8, 2005 at 1:41 pm #362715nicki
ParticipantGary- Are you on 10.4.2? Oh, and have you got it working yet?
I found information in the doc Command Line Administration as well as high Availability. It talks about the scripts to use right after configuring the hostconfig file. Some information is different- for example, the email notification is supposed to be sent from secondary rather than primary server, which makes tons of sense considering it will be down. but there might be something else in there.
August 8, 2005 at 2:44 pm #362717nicki
ParticipantAnother thing: I don’t use for the main server’s hostconfig file the address ending in .255 for the secondary server b/c I couldn’t get the priomary server to come up on the web that way.
The docs tell you to disconnect the primary server after setting it up and reconnect it after you set up the secondary server- this causes it to send me an alert when i plug back in that the ip address is already in use? so, by taking it off the network, the secondary server has already picked up it’s ip address. Wouldn’t that make it primary already before the process even starts?
I tried the script Test and it didn’t help. I copied /usr/bin/false into the new dir and gave it 755.
Nicki
ALso- can you tell me how you set up DNS? maybe that is my problem.August 9, 2005 at 6:50 pm #362744nicki
ParticipantSo, it turns out everything is working fine and I don’t actually think you need the Test script. I just called Apple for support and the way you test it is if both of the network interfaces are down. That’s not really what we were looking for and now I understand why no one is using this. It’s not that useful. The machine has to fail in a major way for the other one to take over. It’s a nice piece of a greater procedure though- you can always write a program to force it to let go of it’s network interfaces under certain circumstances.
-
AuthorPosts
- You must be logged in to reply to this topic.
‘)
Comments are closed