« Previous | Next » 

Revision c4ce868e


Added by Christos Stavrakakis over 10 years ago

Fix bug in snf-dispatcher about ignored messages

Update backendtime only for jobs that have been successfully completed,
since only those jobs update the state of the VM. Else a "race
condition" may occur when a successful job (e.g. OP_INSTANCE_REMOVE)
completes before an error job for the same VM and messages arrive in
reversed order.

Also do not delete VMs when an error OP_INSTANCE_REMOVE arrives.
Delete them only if the operstate of the VM is ERROR.


  • added
  • modified
  • copied
  • renamed
  • deleted

View differences