2019-12-20 - System Operations and Management SIG Agenda and Notes

Q4 will not be released before the beginning of the year. Expect to have fixes made by January 3, deploy the fixes for bug testing on January 6, then release on January 10.

There is no release build.

BugFest bug count was 60 defects. Some were minor. The top-priority bugs were fixed; second-priority in progress. If you download the Q4 branch, you won’t necessarily get all the fixes. 

It's very difficult to track what is in a release due to currently relying on manual process to tie Git to JIRA.  Tools exist, but it would be ideal to move from JIRA to Git for issue tracking.

Jo: It is not practicable that we can work on one ticket only after the other has been closed.

Robbie : At Cornell, github interacts with Jira. It's a setup process.

Ian: We still have to have both levels.

There seems to be a degree of confusion about versioning of modules.  Each module has it's own version.  A spreadsheet for Q4 release is being used to track all the versions!

It's an x.y.z version.

Github has a JIRA integration app that allows tickets to be closed when commits happen.

Need to have a discussion at Wolfcon to try and resolve this. Harry can 90% get it in the schedule. - Anton will make bullet points for this discussion. We will just need three bullet points.

Anton: There are two models of support:

1. vendor-support implementation
2. self-implementation

Ian: The vendor is part of the community.

Dale: Is there a spreadsheet for a list of things that is going into the WolfCon agenda ?
Draft FOLIO WolfCon Sessions: https://docs.google.com/spreadsheets/d/1ngiDnQ9FMy8L3QrD98W4mR9MJ3Elwz0q8MKS4Z0LSOc/edit#gid=694474769

Harry demonstrated the Chalmers "Magic button" service.  Code is available in Github: https://github.com/ChalmersLibrary/chami

Harry: Chalmers are not Cataloguing. They transform LIBRIS records into JSON and automatically load those into FOLIO.
They use OAI-PMH for this. It should work also for others.
Periodically, this "Chami" tool will look if any other other holdings and items has changed, then they take it and transform it to Folio.
It is kind of scripting a batch export at the Union Catalogue level. It is Chalmi middleware.
This is the "magic button" that we have been looking for.
They are doing that open Source and will make it availbale to anyone who will use it.

There is no reason why not connect it to OCLC.
FOLIO is not tied to MARC !

Harry demonstrated the Chalmers FOLIO Stats dashboard.  Code is available in Github: github.com/FOLIO-FSE/folio_stats

There is a dashboard for statistics.
It's a simple cron tab which transfers the data to a flat file.
folio-libris-rtac.azurewebsites.net/statistics/#
You get up-to-the minutes statistics
A flat file is an awesome way to go.
The whole thing is written in Python.

Jo Drexl created a single and dual server postgress cluster as well as a master slave setup.  Postgres has a new feature called streaming replication.  With this there is no time gap between master and slave. The master pipes a write command and pipes that in the same second to the slave. The master waits for the slave to finish. Can happen in sync or async.  When master fails, with a simple file change, slave can be elevated to master status. Multiple slaves can be configured.

So you can do 2 setups with the slave. The slave can work in the warm standby mode. You just have to re-work the IP routing.
Or you can work in a hot standy mode. Then the slave will do read-only queries. You can do db dumps on the slave without affecting the master.
How bullet proof is the master in this master-slave environment ?
It is easy to put a slave in this environment. You can have streaming replication with synchronous slave connections. You can also have two slaves synchronous + 2 slaves asynchronous. There is multiple ways to do this.  You don't have to restart the master.

Ingolf: Happy Holidays & See you next year !