Libserv64 is tricky to triage files off there, as much content is on the Isilon and/or Figgy, in different formats and different levels of description. Seems like a good candidate for a Starfish pilot.
Is it definite that PRDS is going with Starfish? They are definitely doing a pilot, but they are doing a pilot of other things as well. Starfish seems like a good candidate to add our content from Libserv64 to the pilot.
Christa Cleeton thought there was content on Libserv64 but it was all on the Isilon in her case.
Are there audio files on there? There were some that Esmé couldn't find anywhere on the Isilon and copied over from Libserv64 to the Isilon, but it's possible those files were already there and just very difficult to find.
Libserv64 was used by Special Collections pre-PUDL and pre-Isilon. Much of the content got ingested into the Isilon and/or PUDL and then subsequently into Figgy. Content got copied onto the Isilon and then reorganized there. There are still sets of files that are marked "delete" but it is unclear in general if there are any files on Libserv64 that need to be saved.
Starfish would help us eliminate the duplicates and allow us to review just what is left.
This sounds viable; unclear when the Starfish pilot is going to happen. Stephanie will ask at the next Tiger Data meeting about this. If not, Stephanie will reach out to our vendors and see if there is another product we can use.
PostgreSQL 12 - physical cluster (Orangelight and Figgy are the largest applications still using)
This is high priority for Operations and DRDS. We would like to use these resources to prop up production VMWare.
This would really help free up space.
We can sunset this after Orangelight and Figgy (there are small applications still using; would be trivial to remove).
Solr 7 (DPUL, Figgy, Pulmap)
DPUL, Figgy, and Pulmap are the last applications on Solr 7, after which we can sunset.
Once upgraded, we can decommission or repurpose the physical boxes for Solr 7 (libsolr1, libsolr2, and libsolr3).
Machines OPS is decommissioning:
lib-smoothwall
lib-vdp01, lib-vdp02
lib-dom0, lib-dom1
lib-aspacedev1
lib-aeon-dev
lib-ansTest
lib-elixir-test1
Library-Dev
LibRuby-Dev
libserv61-test
puldam-dev
This comes from the recent inventory and call for feedback spreadsheets for moving servers to low-availability hardware.
All of these have been disconnected from the network by now. Some have been archived, as they have been off the network for 30 days or more. The last ones are set to be decommissioned on July 12.
Kate posted the sunsetting evaluation to Confluence.
No big issues identified in the case study.
We need to determine an actual date for decommissioning. Sunsetting Group should revisit this in the fall. PULFA is still used as a standard of comparison for Special Collections, library developers, and researchers for troubleshooting search results in the new application.
No sunsetting date for PULFA yet.
No sunsetting date for PUDL yet.
Tentatively October sunset date for Voyager and SFX.
Meridian could be sunset any time; best to do it at the same time as Voyager and SFX.
Kevin will confirm with Joe M. that there is no outstanding need for this; if not, we can disconnect it from the network soon, before October.