yahoo-eng-team team mailing list archive

[Peter Feiner <peter@xxxxxxxxxxxxxx>]

Nevermind, this was my client-side bug. I was coalescing 'nova list'
requests from different threads without regard to other requests those
threads may have made to nova-api (e.g., boot).

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

-- 
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/1253193

Title:
  instances occasionally do not exist in `nova list` immediately after
  boot

Status in OpenStack Compute (Nova):
  Invalid

Bug description:
  The "boot" and "list" commands in nova don't seem to be sequentially
  consistent. Doing performance testing on the latest openstack code, I
  occasionally observe that the instance is not in the output of "nova
  list". If I re-request the list a moment later, the instance is almost
  always there (i.e., "show" and "list" are eventually consistent).

  Note that above, when I saw "nova ...", I'm not using the command-line
  nova client tool. Instead, I'm issuing the requests from the same
  process with the same novaclient.client.Client instance. The delay
  between invocations of the nova command-line tool would hide the race
  I'm observing.

  I observe this problem fairly rarely; roughly once per 100 instances
  booted.

  My suspicion is that the inconsistency arises from how nova-api
  interacts with the database. Thus it's pertinent to note that I'm
  running with 20 osapi workers.

  I'm using mysql.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1253193/+subscriptions