a specific ip always triggers frozen instances

asked 2019-09-25 13:33:14 -0500

I use openstack horizon to instanciate some vms (ubuntu/bionic). Since yesterday, I experience a very weird behavior. Indeed, one very specific ip (192.168.13.201) always triggers a frozen instance. By frozen, I mean not ping-able and not ssh-able in spite of correctly set security groups (icmp and ssh). Until yesterday, some instances could be created with this ip without any trouble. I tried to rebuild, soft/hard reboot, the instance, attach/detach interface but nothing worked so far. It seems that this ip is sort of definitely corrupted. Is that possible and if so how can I recover from this ?

edit retag flag offensive close merge delete

Comments

Is there any other device in the network that has this IP address? If yes, you should be able to ping it when no instance with this address is up.

Another approach is tracing network traffic with tcpdump or wireshark.

You can also check ARP caches. Perhaps the MAC address(es) provide(s) a clue.

Bernd Bausch gravatar imageBernd Bausch ( 2019-09-25 18:53:54 -0500 )edit

Can you enter the console of a "frozen" instance and look around? Did it get the address from DHCP? Does the DHCP server (dnsmasq, usually) have this address in its leases file?

Bernd Bausch gravatar imageBernd Bausch ( 2019-09-25 18:55:36 -0500 )edit

I deleted the frozen instance with 192.168.13.201 and ping that ip after. I still get Destination Host Uneachable which indicates that there was no other instance with the same ip. Unfortunately I can not enter the frozen instance because it is a ubuntu cloud instance which prevents from login.

eric pellegrini gravatar imageeric pellegrini ( 2019-09-26 01:53:21 -0500 )edit

You can launch a Cirros instance with this address and enter its console. You can also use the Ubuntu instance's console log to see if it gets the IP address.

One question I forgot to ask: Are the instances connected to a self-service network? Is this a floating IP address?

Bernd Bausch gravatar imageBernd Bausch ( 2019-09-26 03:43:57 -0500 )edit

I do not really know about this because I am quite new to these concepts. The only thing I can say is that I do not select any ip in advance. When instantiating a vm, openstack automatically proposes an ip based on the available ones.

eric pellegrini gravatar imageeric pellegrini ( 2019-09-26 04:04:00 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
0

answered 2019-09-26 12:00:35 -0500

The problem came from the fact that there were two entries in the dhcp host file corresponding to my failing ip address. The problem could be solved by killing dnsmasq process then restarting dhcp agent on the network node and finally rebooting the frozen machine.

edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2019-09-25 13:31:12 -0500

Seen: 31 times

Last updated: Sep 25 '19