Can not ping instances when nova-network is restarted

Bug #1005182 reported by Jiang Yong
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
OpenStack Compute (nova)
Incomplete
Undecided
Unassigned

Bug Description

I run openstack essex on CentOS 6.2. I run keystone, glance-api, glance-registry, nova-api, nova-network, nova-scheduler, nova-cert, nova-objectstore and nova-compute on my control node and only nova-compute on a compute node. Before I kill nova-scheduler and restart it, I can use the command ping on my control node to ping instances on control node and compute node. But after I kill and restart nova-scheduler I can not ping instances on control node and compute node, I get the result as below:

[root@cc ~]# ping 10.0.0.2
PING 10.0.0.2 (10.0.0.2) 56(84) bytes of data.
From 10.0.0.1 icmp_seq=2 Destination Host Unreachable
From 10.0.0.1 icmp_seq=3 Destination Host Unreachable
From 10.0.0.1 icmp_seq=4 Destination Host Unreachable
From 10.0.0.1 icmp_seq=6 Destination Host Unreachable

Event I restart all nova compents on both control node and compute node and set two options below:
resume_guests_state_on_host_boot=True
start_guests_on_host_boot=True
let all instances reboot when nova-compute restart, but it does not work.

So I reboot my control node computer and start all services, and restart nova-compute on compute node. Now it works!

Yaguang Tang (heut2008)
Changed in nova:
status: New → Incomplete
Thierry Carrez (ttx)
Changed in nova:
status: Incomplete → New
Revision history for this message
Yaguang Tang (heut2008) wrote :

nova-scheduler has nothing to do with instance network connectivity, can this be reproduced ?

Revision history for this message
Jiang Yong (jiangy) wrote :

Yes. I'm also feel strange, but I have test nova componts one by one and fanally found it is nova-scheduler's problem. If I restart nova-scheduler, I must restart my control node system and then all instances to connect them. I only test private network not including floating ip.

Revision history for this message
Jiang Yong (jiangy) wrote :

I find I was wrong. It's the problem of nova-network. After restarting nova-network, I find it does not restart dnsmasq server and after that dnsmasq server can not discover dhcp request. If I kill the two processes of dnsmasq and restart nova-network. It will work well. Is there any solutions?

Revision history for this message
Russell Bryant (russellb) wrote :

This looks like it is the same as bug 1000853.

What is the version of nova that you have? Where did you get the packages you are using? Hopefully you just need to update to a version that has the fix from that bug in it.

Changed in nova:
status: New → Incomplete
summary: - Can not ping instances when nova-scheduler restart
+ Can not ping instances when nova-network is restarted
Revision history for this message
Jiang Yong (jiangy) wrote :

Hi, Russell, thank you. It is real the same as bug 1000853 and I have solved it now.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.