heat returns wrong stack status on /stack/stack_id call

[Joanna Rozestwińska]

Heat returns wrong stack status on /stack/stack_id call ;

Heat stack-list shows stack's status as 'CREATE_COMPLETED', although REST call responds with 'CREATE_IN_PROGRESS'.
It happens in ~50% of cases, even if stack changed it's status a minute earlier from CREATE_IN_PROGRESS to CREATE_COMPLETED (on heat stack-list called from console).

Heat installed from yesterday's (21th feb 2013) sources.

[Steven Hardy (shardy)]

I cannot reproduce using latest heat master (7149a817b023b5f731a79901af8de4ca671a0247), and python-heatclient 0.2.1

Can the reporter please confirm if they are still experiencing this issue, and if so please provide a step-by-step reproduce procedure?

[Zane Bitter (zaneb)]

Is there any chance that there were multiple instances of the engine running? I have seen a similar problem before where there were two instances of heat-engine running with different data, and requests were handled by each engine alternately.

[Joanna Rozestwińska (joannarozes)]

Unfortunately yes, we still do experience this.

1. We create stack on machine with heat working on it.
2. We wait ~20 seconds (~ because we've tried with many configurations of this time, from 5s to 60s)
3. We ask for stack data on v1/tenant_id/stacks/some_stack_name/some_stack_id
4. In 5 second period we ask for stack data on v1/tenant_id/stacks/some_stack_name/some_stack_id (the timeout is 60s-120s depending on current settings) until it returns 'CREATE_COMPLETED'
5. If it returned 'CREATE_COMPLETED' we ask for stack data one more time.

Resonse from 3rd point says the stack status is 'CREATE_IN_PROGRESS' which is right, as heat stack-list on the machine says the same thing.

Resonse from 4th/5th point also says the stack status is 'CREATE_IN_PROGRESS' which is not right, as heat stack-list on the machine says 'CREATE_COMPLETED'.
Usually stack is created after about 20seconds, so in 4th point we can request for stack-data even 8-20 times and usually one of them is right.

As I wrote earlier, it happens randomly in ~50% of cases.

[Joanna Rozestwińska (joannarozes)]

@Zane There is no such possibility that there are 2 engines running simultaneously.

[Joanna Rozestwińska (joannarozes)]

It occurs that this bug comes from some other library that we using.

Thank you for your help. I am sorry for wasting your time.