[Pacemaker] After a bug is closed, I was confused with default-resource-stickiness

[Adrian Chapela]

Hello,

after the bug =

http://developerbugs.linux-foundation.org/show_bug.cgi?id=3D1852 is =

closed. Now I am confused with the behaviour of default-resource-stickiness.

I want two nodes, one of them as Master (always the same..) but if this =

master fails I need the "Failover Master" remains as master until an =

admin do maintenance on the failed node. For this I have the next =

constraints:

<constraints>
        <rsc_order id=3D"MySQL-IP_Group" first=3D"MySQL" =

first-action=3D"promote" then=3D"IP_Group" then-action=3D"start"/>
        <rsc_colocation id=3D"IP_Group-with-MySQL" rsc=3D"IP_Group" =

with-rsc=3D"MySQL" with-rsc-role=3D"Master" score=3D"INFINITY"/>
        <rsc_location id=3D"loca_MySQL_node1" rsc=3D"MySQL">
                <rule id=3D"rule_loc_MySQL_node1" role=3D"Master" score=3D"=
100">
                        <expression id=3D"exp_rule_MySQL_node1" =

attribute=3D"#uname" operation=3D"eq" value=3D"node1"/>
                </rule>
        </rsc_location>
        <rsc_location id=3D"loca_MySQL_node2" rsc=3D"MySQL">
                <rule id=3D"rule_loc_MySQL_node2" role=3D"Master" score=3D"=
50">
                        <expression id=3D"exp_rule_MySQL_node2" =

attribute=3D"#uname" operation=3D"eq" value=3D"node2"/>
                </rule>
        </rsc_location>
        <rsc_location id=3D"mysql-connectivity" rsc=3D"MySQL">
                <rule id=3D"mysql-pingd-prefer-rule" score=3D"-INFINITY" =

role=3D"Master">
                        <expression id=3D"mysql-pingd-prefer" =

attribute=3D"pingd" operation=3D"lt" value=3D"1000"/>
                </rule>
        </rsc_location>
</constraints>

Another rule is failing for me is the pingd rule but I think it is a =

problem with pingd. I have this messages in error log:
....
pingd[2936]: 2008/11/26_12:46:31 info: ping_read: Retrying...
pingd[2936]: 2008/11/26_12:46:31 info: ping_read: Retrying...
...

I have attached status and config xml. I think pingd problem is when I =

unplugged cable or I down the ethernet interface (ifdown eth0). In these =

cases, pingd have an error "ping_read: Retrying..." and it doesn't =

update value but if it can get  hosts pingd update value. For me this is =

a problem because you could have a switch power failure and this network =

card will go down.

Regards.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: state.xml
Type: text/xml
Size: 20682 bytes
Desc: not available
Url : http://list.clusterlabs.org/pipermail/pacemaker/attachments/20081126/=
60bd824e/state.bin