Cuffdiff 2.2.1.5 SQLite option results in job failures

[jennaj]

This is the most current version on the tool shed and it always fails when the SQLite option is used.

History of the problem as it presented at https://usegalaxy.org: galaxyproject/usegalaxy-playbook#50

What we are telling users to workaround the problem (for at least the last 4 months): https://biostar.usegalaxy.org/p/26894/#27025. In short, to use the earlier version: 2.2.1.3.

Merged duplicated Biostars posts into this one, use as the master for reported problems and final resolution: https://biostar.usegalaxy.org/p/26894/

ping @natefoo @martenson @davebx

[davebx]

I'm investigating this now, should have something to report within 24 hours.

[davebx]

24 hours not counting weekend, that is.

I've tested cuffdiff 2.2.1.5 locally, and it happily outputs SQLite which renders fine with the updated cummeRbund wrapper. I'm going to test with the user's data and see if there's something special in that situation.

[davebx]

Update: the user's data ran successfully on my local environment and failed on main, but I had tweaked my environment previously to get the R script to run. My suspicion is that those tweaks need to be applied to the bioconductor-cummerbund bioconda package, and then that bioconda package needs to be updated on main.

[jennaj]

@davebx -- Identified the root, awesome! Do you want to create a ticket or should I? If ticket/PR/fix already exists, could you link it back in here?

Then close this one out and track other places (usegalaxy-playbook) for main (org).

Now, other usegalaxy* servers should be made be aware and update, too, unless already configured around this issue.

Scope creep: We don't have a regular way to do alert/track important usegalaxy* mods/updates yet, afaik. We need a process for that, in GH if possible... as we build toward coordinating the public sites. Can discuss as a distinct issue -- as part of overall team/community coordination.

ping @bgruening -- input/feedback? :)

[bgruening]

At some point it would be nice to have a usegalaxy.* mailing list for all the admins where we can announce such updates or crucial fixes. usegalaxy.* mailing list should also be part of the security mailing list then.

[abretaud]

Is there a fix for this? I think we have the same problem on galaxy.genouest.org

[jennaj]

Retesting at usegalaxy.org here (still running): https://usegalaxy.org/u/jen/h/test-history-cuffdiff-2215

Will do the same at usegalaxy.eu and link back test history

[jennaj]

Still fails at .org

Cuffdiff is deprecated at eu, but still running test (in progress). Expecting it to also fail with sqlite output, but we'll see. https://usegalaxy.eu/u/jenj/h/test-history-cuffdiff-2215

I noticed that EU has an updated version of HISAT2. ORG has Galaxy Version 2.1.0, EU has (newer) Galaxy Version 2.1.0+galaxy2. Ticket at usegalaxy-playbook to update ORG to latest tool version (released late July https://toolshed.g2.bx.psu.edu/view/iuc/hisat2/a26ed87f444c) galaxyproject/usegalaxy-playbook#155

@bgruening @martenson Side issue: A tool panel search at EU for "hisat" or "hisat2" does not find the tool and once manually found (under group "mapping" the lowest tool version is loaded into the center panel. So, a user needs to use the "versions" menu to switch to the lastest. Perhaps both are knowns?

[jennaj]

Interesting -- the tool works at EU with the SQLite option.

We are now back to there being a problem with the configuration at ORG as the root of the problem. galaxyproject/usegalaxy-playbook#50

Maybe this ticket should be closed out and we only track the problem at ORG? And share the solution with @abretaud ?