[OASIS Issue Tracker] (TOSCA-179) Add "timeout" and "retry" keynames to an operation

[OASIS Issues Tracker <workgroup_mailer@lists.oasis-open.org>]

    [ https://issues.oasis-open.org/browse/TOSCA-179?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=39540#comment-39540 ] 

Moshe Elisha commented on TOSCA-179:
------------------------------------

Thank you, Frank for the valueable comments.

1. I think that this is unrelated to the "retry" option - imagine that retry=1. This is a general question for TOSCA - what should we do if an operation returns that it did not succeed?
IMHO, we should fail the entire deployment even if there are no dependencies but this should be discussed in the TC.

2. I agree this is a topic to discuss. This is true for any program / code with a timeout.
As I see it, timeout should be considered as a failure - The user had explicitly set a timeout which means that he knows that the operation might eventually succeed but he is unwilling to wait or knows that long time shows that something is not right.

If a retry was provided - the TOSCA container will re-run the operation - a user that provides both a timeout and a retry will probably perform a cleanup at the beginning of the operation - for example, before actually doing the operation - it will find & kill previous processes that are running.

Most applications have and know the timeout for each operation and by allowing this in TOSCA we are saving the customer from recreating this mechanism in each customer, for each application, for each supported operation language (script / python / ...).

3. This is also true for operations in general and not necessarily for this issue. We should discuss this in the TC.
Personally, I think we can go either way and possibly also adopt the "OpenStack Heat" way of a "--disable-rollback" option in the deploy request.

> Add "timeout" and "retry" keynames to an operation
> --------------------------------------------------
>
>                 Key: TOSCA-179
>                 URL: https://issues.oasis-open.org/browse/TOSCA-179
>             Project: OASIS Topology and Orchestration Specification for Cloud Applications (TOSCA) TC
>          Issue Type: Improvement
>          Components: Profile-YAML
>            Reporter: Moshe Elisha
>            Assignee: Moshe Elisha
>
> 1. Allow the user to define a timeout for a single operation execution. On expiration, the TOSCA container will mark the execution as failed with an indication that the cause was timeout.
> We can choose a default time unit for this value, we can decide of a pattern "20 m" or we can have another keyname "timeout_unit" that will describe it.
> IMHO, I think that seconds is the minimal time unit that will be required and it is a good default.
> 2. Allow the user to define a retry for an operation execution. 
> If an operation fails to execute (exit code != 0 or timeout), the TOSCA container will execute the operation again until reaching the number of retries.



--
This message was sent by Atlassian JIRA
(v6.2.2#6258)