[Pkg-ganeti-devel] Bug#781084: ganeti: watcher errors on non-master nodes

[Dominic Hargreaves]

Package: ganeti
Version: 2.12.0-3
Severity: normal

The watcher is logging the following errors:

2015-03-23 17:10:12,656: ganeti-watcher pid=28424 ERROR Connect timed out
Traceback (most recent call last):
  File "/usr/share/ganeti/2.12/ganeti/watcher/__init__.py", line 902, in Main
    return fn(options)
  File "/usr/share/ganeti/2.12/ganeti/rapi/client.py", line 263, in wrapper
    return fn(*args, **kwargs)
  File "/usr/share/ganeti/2.12/ganeti/watcher/__init__.py", line 692, in _GlobalWatcher
    client = GetLuxiClient(True)
  File "/usr/share/ganeti/2.12/ganeti/watcher/__init__.py", line 598, in GetLuxiClient
    return cli.GetClient()
  File "/usr/share/ganeti/2.12/ganeti/runtime.py", line 265, in GetClient
    client = luxi.Client(address=pathutils.QUERY_SOCKET)
  File "/usr/share/ganeti/2.12/ganeti/luxi.py", line 101, in __init__
    self._InitTransport()
  File "/usr/share/ganeti/2.12/ganeti/rpc/client.py", line 199, in _InitTransport
    timeouts=self.timeouts)
  File "/usr/share/ganeti/2.12/ganeti/rpc/transport.py", line 101, in __init__
    raise errors.TimeoutError("Connect timed out")

I believe this is attempting to talk to /var/run/ganeti/socket/ganeti-query
which is only expected to exist on the master (quite why 'Connect
timed out' rather than 'no such file or directory' I'm not sure..)

GetLuxiClient is, it appears, expected to throw a NotMasterError exception
which is then caught. I'm not sure why it's not in this case.

-- Package-specific info:
Version symlinks:
  /etc/ganeti/share -> /usr/share/ganeti/2.12
  /etc/ganeti/lib -> /usr/lib/ganeti/2.12
Cluster config version: 2.12.0
Address family: IPv4
Enabled hypervisors: kvm
kvm hypervisor parameters:
  spice_ip_version=0
  nic_type=paravirtual
  cpu_threads=0
  use_chroot=False
  vnet_hdr=True
  cpu_sockets=0
  security_model=none
  cpu_cores=0
  vnc_tls=False
  boot_order=disk
  serial_console=True
  spice_playback_compression=True
  user_shutdown=True
  virtio_net_queues=1
  disk_aio=threads
  reboot_behavior=reboot
  spice_use_tls=False
  spice_use_vdagent=True
  migration_port=8102
  vnc_bind_address=127.0.0.1
  spice_tls_ciphers=HIGH:-DES:-3DES:-EXPORT:-ADH
  migration_downtime=30
  use_localtime=False
  vhost_net=False
  cpu_mask=all
  disk_cache=default
  kvm_path=/usr/bin/kvm
  acpi=True
  vnc_x509_verify=False
  migration_bandwidth=32
  disk_type=paravirtual
  migration_mode=live
  kernel_args=ro
  serial_speed=38400
  root_path=/dev/vda1

-- System Information:
Debian Release: 8.0
  APT prefers testing-updates
  APT policy: (500, 'testing-updates'), (500, 'testing')
Architecture: amd64 (x86_64)

Kernel: Linux 3.16.0-4-amd64 (SMP w/48 CPU cores)
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: sysvinit (via /sbin/init)

Versions of packages ganeti depends on:
ii  adduser              3.113+nmu3
ii  ganeti-2.12          2.12.0-3
ii  ganeti-haskell-2.12  2.12.0-3
ii  ganeti-htools-2.12   2.12.0-3
ii  python               2.7.8-4

Versions of packages ganeti recommends:
ii  drbd-utils                   8.9.2~rc1-2
ii  ganeti-instance-debootstrap  0.14-2
ii  ndisc6                       1.0.1-1+b1
ii  qemu-kvm                     1:2.1+dfsg-11

Versions of packages ganeti suggests:
pn  blktap-dkms  <none>
pn  ganeti-doc   <none>
pn  molly-guard  <none>

-- no debconf information