← Back to team overview

openstack team mailing list archive

instance launched in a wrong compute node

 

hi all:

I have 3 compute nodes,but one(node 3) is down:
root@controller:~/vms# nova-manage service list
2012-12-11 15:10:50 DEBUG nova.utils
[req-a103d7d9-265c-4ef4-a11d-1dba1ccbc9e2 None None] backend <module
'nova.db.sqlalchemy.api' from
'/usr/lib/python2.7/dist-packages/nova/db/sqlalchemy/api.pyc'> from
(pid=14904) __get_backend /usr/lib/python2.7/dist-packages/nova/utils.py:494
Binary           Host                                 Zone
Status     State Updated_At
nova-scheduler   controller                           nova
enabled    :-)   2012-12-11 07:10:48
nova-network     controller                           nova
enabled    XXX   2012-11-22 02:05:14
nova-compute     controller                           nova
enabled    XXX   2012-11-21 03:44:34
nova-cert        controller                           nova
enabled    :-)   2012-12-11 07:10:46
nova-consoleauth controller                           nova
enabled    :-)   2012-12-11 07:10:41
nova-volume      controller                           nova
enabled    :-)   2012-12-11 07:10:48
nova-compute     node1                                nova
enabled    :-)   2012-12-11 07:10:42
nova-network     node1                                nova
enabled    XXX   2012-11-22 02:05:31
nova-compute     node2                                nova
enabled    :-)   2012-12-11 07:10:41
nova-network     node2                                nova
enabled    XXX   2012-11-22 02:05:31
*nova-compute     node3                                nova
enabled    XXX   2012-12-07 09:03:06*

When I am going to launch a instance, it often turns out that the task is
launched on the unavailable node so the task fails:

2012-12-11 15:20:10 DEBUG nova.scheduler.filters.retry_filter
[req-5b648097-33c8-426d-aea8-74be196f8f25 5eb9644d73544e04b347666d1156a002
e6621dd241764ddbaf9cd556882c5aa7] Previously tried hosts: [u'node1',
u'node2'].  (host=node3) from (pid=24208) host_passes
/usr/lib/python2.7/dist-packages/nova/scheduler/filters/retry_filter.py:39
2012-12-11 15:20:10 DEBUG nova.scheduler.filters.ram_filter
[req-5b648097-33c8-426d-aea8-74be196f8f25
5eb9644d73544e04b347666d1156a002 *e6621dd241764ddbaf9cd556882c5aa7]
host 'node3': free_ram_mb:476 free_disk_mb:22528 does not have 4096 MB
usable ram, it only has 970.0 MB usable ram. from (pid=24208) host_passes *
/usr/lib/python2.7/dist-packages/nova/scheduler/filters/ram_filter.py:48
2012-12-11 15:20:10 DEBUG nova.scheduler.host_manager
[req-5b648097-33c8-426d-aea8-74be196f8f25 5eb9644d73544e04b347666d1156a002
e6621dd241764ddbaf9cd556882c5aa7] Host filter function <bound method
RamFilter.host_passes of <nova.scheduler.filters.ram_filter.RamFilter
object at 0x43be250>> failed for node3 from (pid=24208) passes_filters
/usr/lib/python2.7/dist-packages/nova/scheduler/host_manager.py:166
2012-12-11 15:20:10 WARNING nova.scheduler.driver
[req-5b648097-33c8-426d-aea8-74be196f8f25 5eb9644d73544e04b347666d1156a002
e6621dd241764ddbaf9cd556882c5aa7] [instance:
557ade83-151c-425c-bf38-2770e25d0450] *Setting instance to ERROR state.*

I do not know why, any suggestion?

p.s. I use ubuntu 12.10 with Openstack 2012.2

Follow ups