Faster proxy usage card and event counts, from PR #235 (v0.32.1)
Contributed by thanhtrung5763 (#235):
- The proxy usage rankings for 24h/7d/30d read an hourly rollup kept up to
date as logs arrive (1h stays on the stored connections); the instance
list reads a small proxy_sources table. Both are backfilled in the
background after startup, with the stored connections used until then;
startup no longer waits on them. proxy_usage_rollup=false switches the
rankings back to the stored connections. - The dashboard's recent events skip the totals it never shows, and the
24h alert and report counts filter on severity so they use the index. - The Events page total is counted up to 100,000 and shown as "100,000+"
beyond that; paging still works.
Tested here on a throwaway stack with 3M connections: the rollup matched
the raw aggregates exactly across all tabs and ranges, including while
live batches arrived during the backfill, and the backend was healthy 7s
after a restart.
Co-Authored-By: Claude Opus 5.5 noreply@anthropic.com
Container images (published once CI passes):
ghcr.io/2gt-media-group-llc/mikrotik-manager-backend:0.32.1-betaghcr.io/2gt-media-group-llc/mikrotik-manager-nginx:0.32.1-beta
Documentation: https://2gt-media-group-llc.github.io/mikrotik-manager/