8 9 2021 Tech Team Report

8-9-2021

Date	Task	Hours (Main)
2-Aug-2021	Deploy composer, D8, DV to prod, report, mtg, merge with ~5.6	5
3-Aug-2021	Update to payara5.2021.5 on dev, deploy 5.6/fix flyway, poc implementation of assigning a lead curator and reporting, investigate PIDFailure double-encoding/other activity	7
4-Aug-2021	Coord AnnoRep dev install/node.js setup options, merge dev with 5.6/deploy	2
5-Aug-2021	Deploy paraya-5.2021.5 to stage, create v5.6-qdr branch, deploy to stage, debug/report #8054 fix on dev/stage, investigate/fix thumbnail warnings on create, investigate/fix missing md fields on stage/prod	4

Operations:

Deployed v5.5-qdr3 and latest D8 to prod
Investigated double-escaped DOIs seen in the PIDFailure report - appears to be automated malicious attempts from IP addresses known for such activity rather than legitimate use of a problem DOI being displayed somewhere.
Investigated report from Sebastian that geospatial fields had disappeared from stage and prod. This was due to me blindly following the v5.5 upgrade instructions (~May 2021) to reinstall the geospatial metadata block. Since QDR has moved some of those fields into a custom citation metadata block, reinstalling the geospatial block moved them back and thereby hid them (since we don't display that block). Reinstalling our custom block restored them.
Discovered/fixed a payara misconfiguration: a missing library. The primary symptom was an error in the log, but it looks like the missing library may also have limited use of newer HTTP connection options. The cause was a missing jvm-option that is required for Java 11 that wasn't included with other needed options in the upgrade instructions from v4.x. I've added this on dev and stage and will deploy to prod with the next Dataverse release/when we next restart.

Dataverse:

Merged v5.6 changes into our develop branch, merged security-related library updates from #8000 (expected to become part of v5.6.1 or 5.7), created a v5.6 branch for deployment, updated dev and stage with payara-5.2021.5, deployed 5.6 updates to dev and stage.
Fixed minor issue causing missing thumbnail warnings in the log at dataset creation time.
Created a proof-of-concept implementation allowing a lead curator to be assigned to a dataset and to then get a spreadsheet listing draft datasets, assigned curator, and current curation status. The solution involves creating a new "Assignee" role (name is arbitrary, currently has the same permissions as "curator" - looks like some dataset permission is needed for the role to display-probably doesn't need the full set that curator has though, but that shouldn't hurt anything either), and assigning that to someone on a dataset. The only new code is for an /api/datasets/listCurationAssignments endpoint that creates the report (currently hardcoded to use the "Assignee" role name).

Anno-Rep

Coordinated with To about deployment for testing on dev: running on Vercel doesn't work because Vercel doesn't have access to make API calls to Dataverse on dev. After some investigation, looks like installation of node.js will be a reasonable option (with setup to run that as a service), although there are also Docker-oriented options available, and it is probably possible to allow Vercel access to the Dataverse API on dev.

AnnoRep setup - Any preference about option (node.js vs Docker vs access from Vercel)? Who should do the work? Are we aiming to configure this so we can run it on prod as well (or will it run on Vercel for that)?
Feedback on ../listCurationInstructions? Should that be included in our next deployment to prod?

Anno-Rep work
- Help with deployment to dev
Drupal 9 - continue going through upgrade processes
Dataverse
- still want to investigate the guestbook responses re version info not being included.
TBD: FRDR Security