[ClusterLabs] HA static route
S0ke
s0ke at protonmail.com
Tue Mar 15 19:52:38 CET 2016
So I switched the resource to use the following:
<primitive class="ocf" id="p_src_eth0DEF" provider="heartbeat" type="Route">
<instance_attributes id="p_src_eth0DEF-instance_attributes">
<nvpair id="p_src_eth0DEF-instance_attributes-device" name="device" value="eth0"/>
<nvpair id="p_src_eth0DEF-instance_attributes-gateway" name="gateway" value="172.16.2.1"/>
<nvpair id="p_src_eth0DEF-instance_attributes-source" name="source" value="172.16.2.8"/>
<nvpair id="p_src_eth0DEF-instance_attributes-destination" name="destination" value="default"/>
When I do a failover from the master to slave. It seems to add a route of 0.0.0.0/0 but still fails to start the p_src_eth0DEF resource with the same error.
> stderr: RTNETLINK answers: File exists
> stderr: ERROR: p_src_eth0DEF Failed to add network route: to default dev eth0 via 172.16.2.1 src 172.16.2.8
> stderr: DEBUG: p_src_eth0DEF start returned 1
If I remove the 0.0.0.0/0 and manually add the route I want in everything is fine. I did also remove the default gateway from the ifcfg-eth0 file as it was using 172.16.2.1 as the default gw. Still no luck.
Am I using this correctly? Or does it need to be RouteUp?
-------- Original Message --------
Subject: Re: [ClusterLabs] HA static route
Local Time: March 14, 2016 11:04 PM
UTC Time: March 15, 2016 4:04 AM
From: s0ke at protonmail.com
To: dennisml at conversis.de
CC: users at clusterlabs.org
So even if the default gateway is set in /etc/sysconfig/network-script/ifcfg-eth* that could cause it?
-------- Original Message --------
Subject: Re: [ClusterLabs] HA static route
Local Time: March 14, 2016 9:52 PM
UTC Time: March 15, 2016 2:52 AM
From: dennisml at conversis.de
To: s0ke at protonmail.com,users at clusterlabs.org
On 15.03.2016 02:25, S0ke wrote:
> Trying to do HA for a static route. The resource is fine on HA1. But when I try to failover to HA2 it does not seem to add the route.
>
> Operation start for p_src_eth0DEF (ocf:heartbeat:Route) returned 1
>> stderr: RTNETLINK answers: File exists
>> stderr: ERROR: p_src_eth0DEF Failed to add network route: to default dev eth0 src 10.10.5.1
>> stderr: DEBUG: p_src_eth0DEF start returned 1
>
> Is there way to overwrite the route? As it seems to be failing because the route exists already?
The question is why does the route exist already? I you want it to be
managed by Pacemaker then you need to remove it from any other startup
scripts that create that route.
Regards,
Dennis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://clusterlabs.org/pipermail/users/attachments/20160315/49a5d752/attachment-0001.html>
More information about the Users
mailing list