semi-persistent project dirs

[mvdbeek]

Would be pretty cool if I could "resume" a galaxy instance with all its files, tools and database etc.
I've been running a few things on an HPC with the recent slurm / galaxy-job-config-init / progress bar enhancements,
and it's nice, but sometime I'm being kicked out of the interactive job as it reaches its time limit.

In that case it'd be neat if I could start the same instance again and use the job cache to resume the work I was just doing.
Generally speaking it's probably neat to store the install db, shed_tools and the shed_tool_conf file in ~/.planemo.

[mvdbeek]

╭────────────────────────────────────────────── Invocation <f0f0856ecc29bf63> ──────────────────────────────────────────────╮
│ Steps ◆ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ◆ 100% ◆ 13/13 scheduled                                                 │
│ Jobs  ◆ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╺━━━━━━━━ ◆  79% ◆ 11/14 terminal                                                  │
│ Job States ◆ 🟢 11 ◆ 🆕 2 ◆ 👟 1 ◆                                                                                        │
╰───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
srun: error: a000: task 0: Killed

boo! 😆

[jmchilton]

I think the profile concept was headed this direction - I hope a lot of the plumbing is there. I was a little surprised when #1512 didn't mention profiles but maybe the profiles only know how to talk to external Galaxies currently.