Filter RamFilter returned 0 host

asked 2015-01-09 18:24:41 -0600

dlambo gravatar image

running icehouse for a while and now I cant launch any instances, i get error "No valid host was found"

I understand that to mean that my compute node didn't have enough "resources"

But here's the thing, it actually does! and it seems to me that resources are not released when instances are terminated.

Here the error, ramfilter is complaining.

I have rebooted controller and compute as well as restarting all nova services.

still no good

anyone?

2015-01-10 10:03:26.073 1236 WARNING nova.scheduler.driver [req-a153da59-be63-4d60-b021-13dbb99071f0 6df1f912533d40ec9a6c341de3291e5b 5 d81fd3b017945a2b546f00ee27ee9d5] [instance: 729f1cca-5124-44cc-81b3-d991fa7833c2] Setting instance to ERROR state. 2015-01-10 10:04:45.557 1236 INFO nova.scheduler.filter_scheduler [req-ce5624a0-bb72-4227-8fe2-0b13cd1f3a56 6df1f912533d40ec9a6c341de32 91e5b 5d81fd3b017945a2b546f00ee27ee9d5] Attempting to build 1 instance(s) uuids: [u'0ad83e42-2bef-4753-93f3-c4f39f085d12'] 2015-01-10 10:04:45.568 1236 INFO nova.filters [req-ce5624a0-bb72-4227-8fe2-0b13cd1f3a56 6df1f912533d40ec9a6c341de3291e5b 5d81fd3b01794 5a2b546f00ee27ee9d5] Filter RamFilter returned 0 hosts

2015-01-10 10:44:25.726 1873 AUDIT nova.compute.resource_tracker [-] Free ram (MB): -5914 2015-01-10 10:44:25.726 1873 AUDIT nova.compute.resource_tracker [-] Free disk (GB): 79 2015-01-10 10:44:25.727 1873 AUDIT nova.compute.resource_tracker [-] Free VCPUS: -6 2015-01-10 10:44:25.747 1873 INFO nova.compute.resource_tracker [-] Compute_service record updated

lab@compute:/var/log/nova$ free total used free shared buffers cached Mem: 12294520 1205936 11088584 4944 48476 361860 -/+ buffers/cache: 795600 11498920 Swap: 12571644 0 12571644

root@compute:/home/lab# nova host-describe compute +---------+----------------------------------+-----+-----------+---------+ | HOST | PROJECT | cpu | memory_mb | disk_gb | +---------+----------------------------------+-----+-----------+---------+ | compute | (total) | 12 | 12006 | 121 | | compute | (used_now) | 18 | 17920 | 42 | | compute | (used_max) | 18 | 17408 | 42 | | compute | 5d81fd3b017945a2b546f00ee27ee9d5 | 6 | 5120 | 10 | | compute | 263e6804cfbe4704a5c9133c8d496461 | 12 | 12288 | 32 | +---------+----------------------------------+-----+-----------+---------+ root@compute:/home/lab#

root@compute:/home/lab# virsh list --all

Id Name State

2 instance-00000050 running 3 instance-00000057 running - instance-00000039 shut off - instance-0000003a shut off - instance-0000003d shut off - instance-0000003f shut off - instance-00000047 shut off

root@controller:/home/lab# nova-manage service list Binary Host Zone Status State Updated_At nova-cert controller internal enabled :-) 2015-01-10 00:14:21 nova-consoleauth controller internal enabled :-) 2015-01-10 00:14:21 nova-scheduler controller internal enabled :-) 2015-01-10 00:14:21 nova-conductor controller internal enabled :-) 2015-01-10 00:14:21 nova-compute compute nova enabled :-) 2015-01-10 00:14:22

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2015-01-11 22:57:00 -0600

If you look,ram filter is returning Free ram (MB): -5914.

I noticed that you have a some instances running and in shut off state. Im not 100% sure but I think any registered instances count against ram usage , shut off or running.

Can you confirm that you are not exceeding available ram on the compute host even with the instances shut off?

There are also 2 Ram scheduler options you can tweak.

ram_allocation_ratio = 1.5  

(FloatOpt) Virtual ram to physical ram allocation ratio which affects all ram filters. This configuration specifies a global ratio for RamFilter. For AggregateRamFilter, it will fall back to this configuration value if no per-aggregate setting found.

ram_weight_multiplier = 1.0

(FloatOpt) Multiplier used for weighing ram. Negative numbers mean to stack vs spread.

The above settings go in to nova.conf on the controller node.

edit flag offensive delete link more

Comments

I used virsh to undefined the shutdown instances, still got the ram error..

Then I deleted from horizon all instances, even though I was nowhere near the compute nodes resource limits.

Then I can now launch instances.

Strange, but gets me out of the current pickle I was in.

thanks..

dlambo gravatar imagedlambo ( 2015-01-12 00:03:14 -0600 )edit

actually the problem persists.

I now have 0 active instances and can not launch an instance.

dlambo gravatar imagedlambo ( 2015-01-14 15:53:00 -0600 )edit

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2015-01-09 18:24:41 -0600

Seen: 1,748 times

Last updated: Jan 11 '15