yahoo-eng-team team mailing list archive

[Andrew Laski <andrew.laski@xxxxxxxxxxxxx>]

Public bug reported:

Example response:

response status..: <Response [400]>
response time....: 0.434140920639
response headers.: {'content-length': '105', 'via': '1.1 Repose (Repose/6.2.1.2)', 'x-compute-request-id': 'req-a5b3b0b3-28b3-4e0e-9fe3-4d77648e9194', 'server': 'Jetty(9.2.z-SNAPS
HOT)', 'date': 'Fri, 10 Apr 2015 14:37:11 GMT, Fri, 10 Apr 2015 14:37:11 GMT', 'content-type': 'application/json; charset=UTF-8'}
response body....: {"badRequest": {"message": "Block Device Mapping is Invalid: failed to get volume 774787.", "code": 400}}

nova-api log:

2015-04-13 19:27:35.503 5797 DEBUG keystoneclient.session
[req-d9791494-b071-47fa-99d8-db2a5b39a930
dbf01adba9b245369ba32a46d93fdf5f 5930474] REQ: curl -g -i --insecure -X
GET https://example.com/v1/59/volumes/None -H "User-Agent: python-
cinderclient" -H "Accept: application/json" -H "X-Auth-Token: <>"
_http_log_request /opt/rackstack/rackstack.228.11/nova/lib/python2.7
/site-packages/keystoneclient/session.py:195

notice the None in the volume URI.

The issue is that when booting from a volume in cells, the volume is
created in the api cell before there is a device name assigned.  Since
the mapping is looked up by device name when a later update_or_create
call is received it can't find the first mapping and creates a new one.
When later actions, like snapshot, look up the block device mappings for
the instance they find one with no volume_id specified and fail.  This
is assuming that the BFV was attempting to create a volume from an image
during the request.  Booting from a pre-existing volume should not have
this issue.

** Affects: nova
     Importance: Medium
     Assignee: Andrew Laski (alaski)
         Status: In Progress

** Changed in: nova
     Assignee: (unassigned) => Andrew Laski (alaski)

** Changed in: nova
   Importance: Undecided => Medium

** Changed in: nova
       Status: New => Triaged

** Changed in: nova
       Status: Triaged => In Progress

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1444128

Title:
  Cells: snapshot of a BFV instance fails due to volume not found

Status in OpenStack Compute (Nova):
  In Progress

Bug description:
  Example response:

  response status..: <Response [400]>
  response time....: 0.434140920639
  response headers.: {'content-length': '105', 'via': '1.1 Repose (Repose/6.2.1.2)', 'x-compute-request-id': 'req-a5b3b0b3-28b3-4e0e-9fe3-4d77648e9194', 'server': 'Jetty(9.2.z-SNAPS
  HOT)', 'date': 'Fri, 10 Apr 2015 14:37:11 GMT, Fri, 10 Apr 2015 14:37:11 GMT', 'content-type': 'application/json; charset=UTF-8'}
  response body....: {"badRequest": {"message": "Block Device Mapping is Invalid: failed to get volume 774787.", "code": 400}}

  nova-api log:

  2015-04-13 19:27:35.503 5797 DEBUG keystoneclient.session
  [req-d9791494-b071-47fa-99d8-db2a5b39a930
  dbf01adba9b245369ba32a46d93fdf5f 5930474] REQ: curl -g -i --insecure
  -X GET https://example.com/v1/59/volumes/None -H "User-Agent: python-
  cinderclient" -H "Accept: application/json" -H "X-Auth-Token: <>"
  _http_log_request /opt/rackstack/rackstack.228.11/nova/lib/python2.7
  /site-packages/keystoneclient/session.py:195

  notice the None in the volume URI.

  The issue is that when booting from a volume in cells, the volume is
  created in the api cell before there is a device name assigned.  Since
  the mapping is looked up by device name when a later update_or_create
  call is received it can't find the first mapping and creates a new
  one.  When later actions, like snapshot, look up the block device
  mappings for the instance they find one with no volume_id specified
  and fail.  This is assuming that the BFV was attempting to create a
  volume from an image during the request.  Booting from a pre-existing
  volume should not have this issue.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1444128/+subscriptions