Ask Your Question
0

No Valid Host

asked 2015-12-17 15:16:28 -0500

Wendy gravatar image

updated 2015-12-19 21:58:35 -0500

Bipin gravatar image

I have successfully created and terminated several instances of different sizes on my new openstack install. I then terminated all the instances and tried to create an XL instance and encountered the "No valid host found" error. It appears to be a disk issue -- except that I have plenty of space on all the hosts. Here's the output from the log:

2015-12-17 15:20:03.465 23833 DEBUG nova.scheduler.filters.disk_filter [req-fa0559d4-dc56-40ca-b32f-d0c17d7416b7 None] (node-2, node-2.XXXXX) ram:31396 disk:81920 io_ops:1 instances:1 does not have 163840 MB usable disk, it only has 81920.0 MB usable disk. host_passes /usr/lib/python2.7/dist-packages/nova/scheduler/filters/disk_filter.py:60

2015-12-17 15:20:03.466 23833 DEBUG nova.scheduler.filters.disk_filter [req-fa0559d4-dc56-40ca-b32f-d0c17d7416b7 None] (node-5, node-5.XXXXX) ram:31396 disk:81920 io_ops:1 instances:1 does not have 163840 MB usable disk, it only has 81920.0 MB usable disk. host_passes /us r/lib/python2.7/dist-packages/nova/scheduler/filters/disk_filter.py:60

2015-12-17 15:20:03.467 23833 DEBUG nova.scheduler.filters.disk_filter [req-fa0559d4-dc56-40ca-b32f-d0c17d7416b7 None] (node-4, node-4.XXXXX) ram:31396 disk:81920 io_ops:1 instances:1 does not have 163840 MB usable disk, it only has 81920.0 MB usable disk. host_passes /usr/lib/python2.7/dist-packages/nova/scheduler/filters/disk_filter.py:60

2015-12-17 15:20:03.468 23833 DEBUG nova.scheduler.filters.disk_filter [req-fa0559d4-dc56-40ca-b32f-d0c17d7416b7 None] (node-3, node-3.XXXXX) ram:31396 disk:81920 io_ops:1 instances:1 does not have 163840 MB usable disk, it only has 81920.0 MB usable disk. host_passes /usr/lib/python2.7/dist-packages/nova/scheduler/filters/disk_filter.py:60

2015-12-17 15:20:03.469 23833 INFO nova.filters [req-fa0559d4-dc56-40ca-b32f-d0c17d7416b7 None] Filter DiskFilter returned 0 hosts

Here are the file system stats of one of the compute nodes (they are all nearly identical)

root@node-3:~# df -h
Filesystem                                    Size      Used       Avail      Use%      Mounted on
/dev/sda3                                     47G      2.6G       42G        6%                /
udev                                               24G        12K       24G        1%              /dev
tmpfs                                             4.8G        4.2M     4.8G       1%             /run
none                                               5.0M        0           5.0M       0%           /run/lock
none                                               24G         0            24G         0% /         run/shm
/dev/mapper/vm-nova                242G    845M      241G       1%            /var/lib/nova
/dev/sda2                                     181M     33M        139M      20%         /boot
cgroup                                            24G          0           24G          0%             /sys/fs/cgroup

I am trying to start an XL instance.  Here are the flavour details for m1.xlarge (from nova flavor-list)

+--------------------------------------+-----------+-----------+------+-----------+------+-------+-------------+-----------+
| ID                                   | Name      | Memory_MB | Disk | Ephemeral | Swap | VCPUs | RXTX_Factor | Is_Public |
+--------------------------------------+-----------+-----------+------+-----------+------+-------+-------------+-----------+
| 5                                    | m1.xlarge | 16384     | 160  | 0         |      | 8     | 1.0         | True      |

Any suggestions would be helpful. Thanks!

Wendy

edit retag flag offensive close merge delete

Comments

Do you use a local disk on each compute node or do you use a shared storage?

Herr-Herner gravatar imageHerr-Herner ( 2015-12-18 01:06:47 -0500 )edit

Local disk.

Wendy gravatar imageWendy ( 2015-12-18 09:12:03 -0500 )edit

nova flavor-show m1.xlarge

nova hypervisor-stats

Looks like, nova scheduler filter doesnt passes the Disk_Filter.

Bipin gravatar imageBipin ( 2015-12-19 22:01:14 -0500 )edit

2 answers

Sort by » oldest newest most voted
0

answered 2015-12-21 23:41:24 -0500

Prateek K gravatar image

Based on what you told we have the following points :-

  • All the instances have been terminated.
  • The Disk of VM can be on compute node.

Solution would be to on the compute nodes that are there in the environment and navigate to '/var/lib/nova/instances/'

Under this directory you would see some folders having big numbers(UUID's of previous instaces). If you are sure there are no instances shown when you do a 'nova list' then remove all the folders that are present under this directory. All your space will be recovered and you will be able to spawn new instances.

edit flag offensive delete link more
0

answered 2015-12-21 07:48:41 -0500

Vinoth gravatar image

Increase the DISK overcommit ratio.

or

Delete all the unwanted Shutoff VM which might block the free space allocated to it.

Thanks,

Vinoth

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2015-12-17 15:16:28 -0500

Seen: 645 times

Last updated: Dec 21 '15