Cluster Controller Forgets Routing at Reboot
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
eucalyptus (Ubuntu) |
Confirmed
|
Undecided
|
Unassigned |
Bug Description
Dear all,
I found a bug in the Cluster Controller package... maybe especially in the iptables package. of Lucid and Maverick. that affect both eucalyptus 1.6.x and 2.0
I'm using the topology CLC,WS3,CC+SC,NC
where the CC+SC has 2 interfaces and does the routing between the 2 networks separating the "CLC,WS3" network and the "NC" network.
From the NC, I can't ping anything else but other NCs (while I'm supposed to be able to)..
and to fix that I had to reapply to forgotten iptables rules (that are actually cleared at each reboot for some reason)
So I created a script that is executed at each boot
"
iptables --table nat --flush
iptables --delete-chain
iptables --table nat --delete-chain
iptables --table nat --append POSTROUTING --out-interface eth0 -j MASQUERADE #Public Interface
iptables --append FORWARD --in-interface eth1 -j ACCEPT
iptables-save > $EUCALYPTUS/
service eucalyptus-cc restart CLEAN=1
service eucalyptus-sc restart
"
but in other distros this doesn't happen with Eucalyptus...
Based on the information, i am marking as confirmed. This requires further work to determine a suitable remedy.