Bug#539303: ganeti2: gnt-cluster init does not work

Sure, i will give further infomation tomorrow or on Monday. (No testsetup at home)

2009/7/30 Iustin Pop <iusty@xxxxxxxxx>

On Thu, Jul 30, 2009 at 04:27:14PM +0200, Maximilian Mill wrote:
> Package: ganeti2
> Version: 2.0.1-2
> Severity: grave
> Justification: renders package unusable
>
> Hello,

Hi and thanks for the report.

> it looks like 'gnt-cluster init --vg-name=vg0 cluster.xyz.com' doesn't create
> all needed files for the cluster to work. ll /var/lib/ganeti:
> insgesamt 20
> -rw------- 1 root root 2422 30. Jul 13:29 config.data
> -rw------- 1 root root  416 30. Jul 13:29 known_hosts
> drwx------ 3 root root 4096 30. Jul 13:30 queue/
> -r-------- 1 root root 1880 30. Jul 13:29 rapi.pem
> -r-------- 1 root root 1880 30. Jul 13:29 server.pem
> -rwxr-xr-x 1 root root    0 30. Jul 13:30 watcher.data*

This means that the initial part of the init worked, but ganeti-masterd
didn't start correctly.

> I found this problem when i was trying to add a node. This error message comes:
> gnt-node add anc
>         Corrupt configuration file: Can't read from the ssconf file: '[Errno 2] No
>         such file or directory: '/var/lib/ganeti/ssconf_master_node''
>         Aborting.
>
> I have this bug with 2.0.1-2 from unstable on a lenny maschine(bpo'ed).
>
> I did a repack with the new upstream 2.0.2-1 and everything works like it
> should.
> Now gnt-cluster init created these files in /var/lib/ganeti and everything works:
> insgesamt 56
> -rw------- 1 root root 2422 30. Jul 13:32 config.data
> -rw------- 1 root root  416 30. Jul 13:32 known_hosts
> drwx------ 3 root root 4096 30. Jul 13:32 queue/
> -r-------- 1 root root 1880 30. Jul 13:32 rapi.pem
> -r-------- 1 root root 1880 30. Jul 13:32 server.pem
> -r--r--r-- 1 root root   35 30. Jul 13:32 ssconf_cluster_name
> -r--r--r-- 1 root root    0 30. Jul 13:32 ssconf_cluster_tags
> -r--r--r-- 1 root root   25 30. Jul 13:32 ssconf_file_storage_dir
> -r--r--r-- 1 root root    0 30. Jul 13:32 ssconf_instance_list
> -r--r--r-- 1 root root   40 30. Jul 13:32 ssconf_master_candidates
> -r--r--r-- 1 root root   15 30. Jul 13:32 ssconf_master_ip
> -r--r--r-- 1 root root    8 30. Jul 13:32 ssconf_master_netdev
> -r--r--r-- 1 root root   40 30. Jul 13:32 ssconf_master_node
> -r--r--r-- 1 root root   40 30. Jul 13:32 ssconf_node_list
> -r--r--r-- 1 root root    0 30. Jul 13:32 ssconf_offline_nodes
> -r--r--r-- 1 root root   40 30. Jul 13:32 ssconf_online_nodes
> -r--r--r-- 1 root root    6 30. Jul 13:32 ssconf_release_version

It is strange that 2.0.1 is broken  and 2.0.2 works, since there have
been no fixes between 2.0.1 and 2.0.2 in regard to cluster init.

Are you sure this is not a problem of the backport somehow? The failed
cluster init should have left some error detals in the logs
(/var/log/ganeti/*), can you check that out and give the errors details?

thanks!
iustin