[Pacemaker] eth0:0: warning: name may be invalid
Ty! Boyack
ty at nrel.colostate.edu
Fri May 29 03:37:44 EDT 2009
Interesting. Sorry to have offered things that you already knew :-)
I think you may have a bug case -- I would imagine that systems with two
interfaces on the same subnet are less tested than other scenarios. I
know I had a similar situation on a HA pair for a firewall, and I just
ended up writing my own resource script (old version of heartbeat, not a
true OCF script). You may want to do the same or just copy the IPAddr2
script to your own new ocf agent and modify it. On a quick glance, it
does look like that script might take the 'nic' parameter as some sort
of suggestion, but regardless it runs the 'findif' utility to make some
auto-determination of which interface it should really use. If you are
already sure you want it to be eth0:0, I would just edit that script to
skip the findif parts and force it to take your 'nic' parameter. It
should be a fairly small edit, and it would at least get you going.
If you have not played with adding your own resource scripts before, I
would add them in your own directory parallel to the "heartbeat"
directory. Then just set your "provider" to the name of your
directory. The only reason for this is to prevent an upgrade from
wiping out your changes.
Good luck,
-Ty!
Димитър Бойн wrote:
> Thanks a lot!
>
> Indeed! - I do have two interfaces on the same network... :)
>
> Should this inability be considered a bug?
>
> Best Regards!
>
> ___ Dimitar Boyn
>
> ----- Original Message -----
> From: Ty! Boyack <ty at nrel.colostate.edu>
> To: 'pacemaker at clusterlabs.org' <pacemaker at clusterlabs.org>
> Sent: Thu May 28 23:36:39 2009
> Subject: Re: [Pacemaker] [LIST:Pacemaker] eth0:0: warning: name may be invalid
>
> I might be wrong, but unless you NEED to specify which interface should
> carry your alias, you don't need to specify it at all. If you only
> specify an address, it will try to find an interface that already has an
> address on the same subnet, and apply your new IP as an alias to that
> interface. Of course this will fail if you have multiple interfaces on
> the same subnet, but I don't know if you have that situation.
>
> My config looks like this:
> <primitive class="ocf" type="IPaddr2" provider="heartbeat" id="ipaddr-nfs1">
> <instance_attributes id="ipaddr-nfs1-ip">
> <nvpair id="nfs1-ip-addr" name="ip" value="xxx.yyy.zzz.112"/>
> </instance_attributes>
> </primitive>
>
> It searches for an existing interface where xxx.yyy.zzz.112 would be
> local (in my case that is an interface with an IP of xxx.yyy.zzz.111),
> and adds xxx.yyy.zzz.112 as eth0:0. This works for me, and I've used
> multiple primitive definitions like this and it adds each interface as
> the next alias number (eth0:0, then eth0:1, then eth0:2), but as things
> move around you are not always guaranteed which alias number this
> interface will get.
>
> So it might be worth a try to just remove your nvpair line that deals
> with the nic.
>
> I do have an older version of some of the software, it may have changed.
>
> -Ty!
>
>
>
> Димитър Бойн wrote:
>
>> Thank you, Neil!
>>
>> I have already tried this but eth0 has IP Address "on boot"
>> The resource does not start if I change the "nic" value tp eth0 only :(
>> and I indeed need it to be just additional eth0:0.
>>
>> Any ideas what could be wrong?
>>
>> Thanks!
>> ./Dimitar Boyn
>>
>> -----Original Message-----
>> From: Neil Katin [mailto:linuxha at askneil.com]
>> Sent: Thursday, May 28, 2009 9:29 PM
>> To: pacemaker at clusterlabs.org
>> Subject: Re: [Pacemaker] eth0:0: warning: name may be invalid
>>
>>
>> Use "eth0", not "eth0:0" for the "nic" nvpair.
>>
>> As the script built-in docs say:
>>
>> The base network interface on which the IP address will be brought
>> online.
>>
>> If left empty, the script will try and determine this from the
>> routing table.
>>
>> Do NOT specify an alias interface in the form eth0:1 or anything here;
>> rather, specify the base interface only.
>>
>>
>> Димитър Бойн wrote:
>>
>>
>>> Hi,
>>>
>>> How do I get rid of these type warnings? -
>>>
>>> May 28 18:56:52 c001mlb_node01a lrmd: [18066]: info: RA output:
>>> (ip-c001drbd01a:monitor:stderr) eth0:0: warning: name may be invalid
>>>
>>>
>>>
>>> My resource is configured (and seems to work just fine) by:
>>>
>>> <primitive class="ocf" provider="heartbeat" type="IPaddr2"
>>> id="ip-c001drbd01a">
>>>
>>> <instance_attributes id="ia-ip-c001drbd01a">
>>>
>>> <nvpair id="ia-ip-c001drbd01a-ip" name="ip"
>>> value="192.168.80.213"/>
>>>
>>> <nvpair id="ia-ip-c001drbd01a-nic" name="nic" value="eth0:0"/>
>>>
>>> </instance_attributes>
>>>
>>> <operations>
>>>
>>> <op id="op-ip-c001drbd01a" name="monitor" interval="21s"
>>> timeout="5s"/>
>>>
>>> </operations>
>>>
>>> </primitive>
>>>
>>>
>>>
>>> Thanks!
>>>
>>>
>>>
>>> ./Dimitar Boyn
>>>
>>>
>>>
>>>
>>> ----------------------------------------------------------------------
>>> --
>>>
>>> _______________________________________________
>>> Pacemaker mailing list
>>> Pacemaker at oss.clusterlabs.org
>>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>>
>>>
>> _______________________________________________
>> Pacemaker mailing list
>> Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>> _______________________________________________
>> Pacemaker mailing list
>> Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>>
>>
>
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
>
More information about the Pacemaker
mailing list