End-to-end deployment of V2 Loc-I Cache

CSIRO-enviro-informatics / loci.cat

Latest updates from the Loc-I Project

https://loci.cat

Other

0 stars 1 forks source link

End-to-end deployment of V2 Loc-I Cache #50

Open jyucsiro opened 4 years ago

jyucsiro commented 4 years ago

Version 2 Loc-I Cache features updated ASGS features and ontology descriptions implemented in loci datasets (GNAF, Geofabric, ASGS).

This issue tracks

Additional tasks:

Testing using pyloci to ensure content compatibility in terms of excelerator use cases
Any refactoring of Loc-I components to match updated content (loci-integration-api, loci-integration-explorer-app, jupyter examples, excelerator)

jyucsiro commented 4 years ago

@benjaminleighton @ashleysommer it appears the process for deploying V2 Loc-I Cache is ironed out now (we think we know where the gaps are now and have tested solutions).

Fair to say if we repeat the cache deployment, we should have a working cache as expected? Is that all that there is to do?

benjaminleighton commented 4 years ago

Linksets still need to be patched post deployment to fix the subjectTarget/objectTarget issue. A prospective fix for this is merged but needs better testing through a full linkset regeneration, which we might not have time for right now. There is some uncertainity around exactly how the cache should be configured for optimum performance after it's built. Those are the things I can think of.

jyucsiro commented 4 years ago

Ok, how about we open an issue for the full linkset regeneration (we won't have time to do that this FY i think). Another for investigating performance tuning of the cache. Both issues can be deferred to be considered in Loc-I Phase 2 (i.e. next FY).

In the meantime, if we need to redeploy the cache to prod, we can live with doing a patch for Linksets post deployment to fix the subjectTarget/objectTarget issue?

jyucsiro commented 4 years ago

FYI Linkset patching https://github.com/CSIRO-enviro-informatics/loci-cache-scripts/issues/43

benjaminleighton commented 4 years ago

New issues for linkset regen and performance of the cache are a good idea. On performance maybe that issue should be big enough to allow investigation of alternative backend dbs. Patching is fine with me as long as someone remembers.