I’m trying to do some pipeline-y kind of integration for deadline in submitting jobs en-masse, and was wondering if there was any better way to leverage deadlinecommand to get back the resulting JobIDs beyond capturing stdout and having to parse what comes back to bind jobIDs back to the jobs i’m submitting for a report and a downstream process that would like to set up another wave of pending jobs on those JobIDs
It’s not horrible but feels a weird enough that I’m doing a "^(.)=(.)$` type of regex to look through stdout that it feels like there might be a flag or some other kind of approach i’m missing.
Yeah, 100% that’s something that’s tempting - I’m trying to keep the setup as low-complexity as possible initially.
All the hosts will be guaranteed to have a deadline client on them but it’s also arguable that it might be better to enforce having the web service running in another container alongside the mongo db so the code can communicate from non-farm boxes. I just… don’t want to add in another service that someone needs to know about and to maintain.
Yeah, because i’m doing batch work with this right now, the processing time is like, orders on orders of magnitude worse than any speed i can get from a different method because it’s not interactive. As the tooling becomes more artist-side then it’ll be the push to change it but til then maintenance issues are the biggest potential costs.
