Re: [openstack-dev] [mistral] Using mistral to start a not-to-die task

2017-03-24 Thread Sridhar Ramaswamy
Renat: Perfect. A combination of 1 and 2a would fit our needs. One reason we are venturing toward an async workflow is to address the scale limitation in tacker. Now, do you've some pointers how well mistral scales (so that we don't just move our scale problem down to the mistral dependency).

Re: [openstack-dev] [mistral] Using mistral to start a not-to-die task

2017-03-23 Thread Renat Akhmerov
Another option: 1. Run a workflow with a loop (i.e. task calls itself), in order to have a delay between task executions you can use either wait-before/wait-after task policies 2.a The workflow can be stopped via Mistral API by a 3rd party, if needed. 2.b The workflow can have a special task

Re: [openstack-dev] [mistral] Using mistral to start a not-to-die task

2017-03-22 Thread Lingxian Kong
Yeah, as Bob said, cron-trigger is what you are looking for. As our discussion on IRC, currently, Mistral doesn't support to execute shell command directly, my suggestion is, you could consider using http action (which is supproted by Mistral out of the box) or providing a host (physical or

Re: [openstack-dev] [mistral] Using mistral to start a not-to-die task

2017-03-22 Thread HADDLETON, Robert W (Bob)
Hi Gongysh: You are looking for mistral cron-triggers. See https://docs.openstack.org/developer/mistral/terminology/cron_triggers.html Bob On 3/22/2017 7:16 PM, gongys2017 wrote: Hi mistral stackers, Tacker is using the mistral as its part of system. Now we have a requirement: tacker

[openstack-dev] [mistral] Using mistral to start a not-to-die task

2017-03-22 Thread gongys2017
Hi mistral stackers, Tacker is using the mistral as its part of system. Now we have a requirement: tacker server registers an openstack as its NFVI, and needs to ping__ OpenStack Development Mailing List (not for usage