cyberagent / patriot-workflow-scheduler Goto Github PK

add /worker/status to check worker status to replace /worker with worker configuration facilities.

Error wording does not appear when I start a user does not have write access to the log directory

Search API itself is working fine, but the web client doesn't pass the input parameter to the API, which results in fetching all jobs.

https://gist.github.com/tanak/a5a742c68982ffdcf25c

POST methods, both job_api_servlet's and job_servlet's, accept a request that generates an existing job_id in the database.

At least, job_api_servlet's POST method should be splitted into smaller APIS to take backward compatibility into consideration.

BigQuery has "configuration.dryRun" option to execute dry-run.
I would like load_to_bigquery to accept this option when -t parmeter is given.

https://cloud.google.com/bigquery/docs/reference/v2/jobs

It would be nice if you can get how many results there are when searching.

Currently, only paths of deprecated naming convention are included in "register/execute" commands with directories.
Set default mode of the get_batch_files to all for including all files in default.

http://cyberagent.github.io/patriot-workflow-scheduler/cli.html
https://github.com/CyberAgent/patriot-workflow-scheduler/blob/develop/lib/patriot/util/script.rb#L18

Workers does not commit completion status of jobs if the jobs are updated.
The 'register' tool with the --keep-state option updates jobs configuration other than state, and cause to workers not to update the state.
In case of the updated jobs in the running state, the jobs remain at the running state even if they are finished.

Due to a9392a8, custom commands with hash attributes are broken since they expect keys are string.

sh{
    start_after '23:50:00'
    name "test"    
   commands "hoge hoge'"
}

start_after is not work.
start_after value is not registered in the database. Start moving immediately job.
Cause is following,
http://dba.stackexchange.com/questions/48704/mysql-5-6-datetime-incorrect-datetime-value-2013-08-25t1700000000-with-er

When you search something and click "next", you will get no result.

1. search something on some page like "Waiting" in web interface
   The results should be more than 50
2. click "next"
   you will get no result
3. click "search"
   you will get the results as same as No.1

JobViewer with tilt2 requires gems of which dependency is not configured, and therefore, it does not work.
The version of tilt should be explicitly set in the gemspec.

I think load_to_bigquery should continue to process even after dealing data with bad records less than maxBadRecords setting.

There is no title set on html head, so when you bookmark a page, the name of the bookmark page will be always set as "Home".
Preferably, the title of http://.../job/list/4 should be
Patriot Workflow Scheduler - Failed Jobs
or something.

GCP configuration files are only needed on job execution and not needed in register phase (which would be performed on a different host).

https://github.com/CyberAgent/patriot-workflow-scheduler/blob/develop/plugins/patriot-gcp/lib/patriot_gcp/command/load_to_bigquery.rb#L20

https://bugs.ruby-lang.org/issues/7917