[
https://issues.apache.org/jira/browse/AIRAVATA-3549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17459558#comment-17459558
]
Marcus Christie commented on AIRAVATA-3549:
-------------------------------------------
Design of Airavata API server changes:
Airavata API
- void fetchIntermediateOutput(AuthzToken authzToken, String experimentId,
List<String> outputs)
- publish message to experiment channel with MessageType.INTERMEDIATE_OUTPUTS
- ExperimentIntermediateOutputsEvent(experimentId, gatewayId, List<String>
outputNames)
- OSH.ExperimentHandler.onMessage
- in handling, check that there isn't already an active OUTPUT_FETCHING task
for the file(s)
- we might want a timeout of the OUTPUT_FETCHING task, so we aren't waiting
forever
- create a process, but there are no inputs and only outputs are the
specified ones
- also check that experiment isn't already completed
- create the output data staging tasks, task type OUTPUT_FETCHING
- publish a PROCESS_LAUNCH event
Helix changes:
- when getting PROCESS_LAUNCH event, check if it only contains OUTPUT_FETCHING
tasks. If so, run a new workflow that fetches those output files.
> Fetch intermediate output files
> -------------------------------
>
> Key: AIRAVATA-3549
> URL: https://issues.apache.org/jira/browse/AIRAVATA-3549
> Project: Airavata
> Issue Type: New Feature
> Components: Airavata API, Django Portal, helix implementation
> Reporter: Marcus Christie
> Priority: Major
>
> Add fetching of intermediate files for running experiments to the Airavata
> API. The main use case is for API users to get the output of log files so
> that they can determine how the experiment is going. An API user may want to
> cancel a computational job that isn't converging, for example.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)