Skip to content
GitLab

[#101] Optimize GRAPPA calls, enhance cache, bug fixes batch

Merged Emmanuel Ormancey requested to merge 101-optimize-grappa-calls-and-cache into master

Closes #101 (closed)

  • Added parameters to allow config of Authz Service API (to use QA for example
  • Added Cache TTL parameter
  • Now use /precomputed for Grappa GetMembers
  • Optimized Channel.isAdmin to reduce GetMembers calls
  • Added pg_stats_statements module to postgres for local deployments in docker-compose-service-shared.yml to facilitate query perf tuning (https://www.geekytidbits.com/performance-tuning-postgres/)
    • connect pg and run : CREATE EXTENSION pg_stat_statements;
    • then SELECT * FROM pg_stat_statements ORDER BY total_exec_time DESC;
      • if you need to reset the data SELECT pg_stat_statements_reset();
    • then get the long query and EXPLAIN ANALYSE paste_the_query_here
    • psql tips:
      • Show timing of queries: \timing
      • enable disable pages results: \pset pager
  • Replace Cache module by Memoizee module (https://www.npmjs.com/package/memoizee) to prevetn too many parralel requests to GRAPPA that kill the service by triggering their anti-flood system
  • Updated typeorm to latest (0.2.38) to address performance issue on home page with lots of channels + members
  • Fixed skip/take bug on long lists slice (convert skip/take to numbers)
  • Improve find-all-channels query. Thanks for Carina we now have a splitted query quite fast.
  • Improved add/remove members

General findings: https://codimd.web.cern.ch/3XokN3PjSeuUXqnRHsKZzg

New ENV:

AUTHORIZATION_SERVICE_API_AUDIENCE=authorization-service-api
AUTHORIZATION_SERVICE_API_TOKEN_URL=https://auth.cern.ch/auth/realms/cern/api-access/token
AUTHORIZATION_SERVICE_API_BASE_URL=https://authorization-service-api.web.cern.ch

# Optional Cache TTL, default 300
CACHE_GRAPPA_TTL=1200
#CACHE_TOKEN_TTL=300
Edited by Emmanuel Ormancey