[Users] Host discovery failing due to host network being lost

[Haim Ateya]

On Jan 28, 2012, at 10:27, Deepak C Shetty <deepakcs at linux.vnet.ibm.com> wrote:

> Hi ovirt users,
> 
> I was trying to check on IRC the below, did not get any replies, so sending this mail.
> Can someone help me understand what could be the issue that causing my managed host to lose network settings, when i try to discover it from ovirt-engine ?
> I have to then manually (using a remote console) re-setup the network.. it even overrites my ifcfg-eth0 file
> 
> I see a new bridge getting added called ovirtmgmt and route entry corr. to that.. the ifcfg-eth0 also has that entry
> I feel that ovirt scripts running on the host should atleast take a backup of the ifcfg* files they modify so that in case of issues like these, user can login via remote console and restore the original ifcfg file.
> 
> This is what i asked on IRC... responses appreciated...
> 
> deepakcs> Hello, i just configured ovirt-engine and discovering my first managed host.. got some Qs
> <deepakcs> what diff does it make when i select "override ip tables" check box during new Host workflow ?
> * djasa has quit (Ping timeout: 480 seconds)
> <deepakcs> this is the second time i am discovering this host and everytime the host discovery hangs during "Installing" and if i check the host, it goes out of network.... has anybody faced this problem before ?
> <deepakcs> While discovering the host, the host network ipv4 address is gone ! and thus the host goes out of network
> 
> Some more updates..
> 1) I manually re-setup the host network, made it pingable, esp from the ovirt-engine node, and clicked on Re-install on the web gui, this time with the 'override ip tables' check enabled
> 2) this time it successfully completed the node bootstrap steps (as seen from the Events window) and status was Reboot
> 3) But after reboot the same thing happened.. the host lost ipv4 addr, route entries are completely gone and onthe web gui the host is seen as non-responsive, which is expected bcos the host is not on the network anymore...
> 
> what is the reason the above is happening ?

Hi, 

First, in order to understand the reason for bootstrap failures, please attach engine.log (ovirt-engine) and node logs (under /tmp/vds_bootstrap.log, vds_installer.log).

Second, what you describe regarding the loss of network could derive from the following reasons:
- host is configured with bonding - but bonding is not configured correctly.
- NetworkManager interfere  with networking config.
- bridge (ovirtmgmt) is not set to with any BOOTPROTO or ONBOOT.

Please attach network scripts (ifcfg-eth0, ifcfg-ovirtmgmt) and /var/log/messages .

Haim
> 
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20120128/05f4c475/attachment-0002.html>