2024-07-22-Global Services Test
Date
Jul 22, 2024 13:00-15:00 UTC
Participants
Rémy GIRAUD (SC-IMT Chair)
Lei XUE (SC-IMT Co-Vice-Chair)
Yoritsugi YUGE (JMA)
Masato FUJIMOTO (JMA)
Steve Olson (NOAA)
Kari SHEETS (NOAA)
Chemseddine El Garai (DMN)
Xinqiang HAN (CMA)
Wenjing GU (CMA)
Schremmer Antje (DWD)
José Mauro DE REZENDE (INMET)
Saad Almajnooni (NCM)
Mohammed ALGHAMDI (NCM)
Tom KRALIDIS (ECCC)
Jeremy TANDY
WMO Secretariat
Anna Milan
Maaike Limper
Timo Proescholdt
Apologies
Max Marno (SYNOPTIC)
Thorsten Büßelberg (DWD)
Hyumin EOM (KMA)
Ara Eom (KMA)
Omar Daftardar (NCM)
Agenda and Notes (DRAFT)
Item | Presenter | Notes |
---|---|---|
Opening |
|
|
Review actions |
| Actions of Meeting on July 8th:
|
GDC testing | Tom | presented tests and steps for GDC tests
There are demo records and test scripts need to be developed (Maaike) regarding 8.4.2.2 “on successful completion” – MQTT is not required for non-real time data (Tom) test data should have real time and non real time data. (Jeremy) would it be worth putting something in the test that accounts for archive/nonRT data (….) (Remy) concern about how the metrics are provided (8.4.2.10) - it should not be a requirement. (Antje) regarding comparing centre-ids (8.4.2.5) - the origin message should be under the right centre-id. (Remy) e.g.. DWD is hosting the node for Israel, DWD is the link but centre-id is related to Israel; (Tom) so in other words, this is not a use case but perhaps this needs to be explicit in the Guide (Remy) at this point I don’t think it needs to be documented (Jeremy) I think it would be worthwhile putting a paragraph in the Guide …”if you are publishing on behalf of another centre… do XZY” (see action 2) (Remy) the GDC is not subscribed to origin, but is subscribed to the GC? see Action 3 (Remy) I think the GC also subscribes to the origin data (Remy) the GC should cache all the records, there is no notion of core vs recommended (Jeremy) 2.7.4.2 of the Guide has clear guidance |
test infrastructure | Remy | There will be 10 virtual machines for testing; (Timo) we will be able to provide these machines; they will be geographically distributed and use the DNS links in the WIS dev domain; It would be good if we can spin them up on demand; (Remy) If the cost is extremely prohibitive, there is no need to have these machines always on during the test period. It would be useful to look at the cost; |
GS testing tools | Remy | two tools identified in Tokyo. Remy has updated the GH with more details: (1) Preparing GB for testing and (2) Activating the tests on fake WIS2 Nodes. (Antje) there should be minimum delay between origin and update? I think there should be a delay. (Remy) there is nothing in the specification about this right now. (Antje) the GC will discard the data if the content is not included, but I think that the GC should try to download (Remy) add embedded to the text (Antje) it would be good to have a manager script to publish tests in the same order […?] (Remy) we can either write one large test or we can define an order; (Timo) there will be tests with different components that need to work together and there will be resets (Remy) all of the test instances will have scenarios and how this is operated(?) and reset is not included (Tom) how do you see this working with GDC testing? (Remy) we can copy the metadata to the fake wis2 node or we put them on GitHub; (Antje) it would be an advantage to have Prometheus inside the wis2dev environment (Remy) agreed; at the moment on the wis2 dev environment, there is Grafana, but not Prometheus however, it’s easy to add.
|
Demo | Remy | (Antje) should the common json files be in a GitHub repo? (Remy) they are there under scenario (Jeremy)… (Maaike)… (Antje) there is a branch gc-tests with ../test/global-cache/scripts/metrics.py and ../scripts/manager-test-py |
|
|
|
Actions:
(Anna) Check the Guide: what is the behavior for GC regarding recommended data?
(Tom) add issue to WIS2 Guide for what to do when publishing on behalf of another centre.
(Tom) verify what the GDC is connected to