fix: cap tap CPU to 0.25 each to prevent DB starvation The two ATProto firehose backfill processes (tap, tap-knots) were consuming 50%+ CPU each, pushing postgres to 85% CPU and causing 118-second page load times. Capping each at 0.25 CPUs brings page loads back to ~1 second while backfill continues at a slower pace.

Author: Aaron Steven White
Commit 167ce82f2cef36f324116e8fb6d66724e6394be9
Parent: c043cb60ae
Structural diff unavailable

These commits were pushed via plain git push, so no pre-parsed schemas are available. Install git-remote-cospan and re-push via panproto:// to see scope-level changes, breaking change detection, and semantic diffs.

brew install panproto/tap/git-remote-cospan
1 file changed +8 -0
@@ -99,6 +99,10 @@ services:
9999     image: ghcr.io/bluesky-social/indigo/tap:latest
100100     container_name: cospan-tap
101101     restart: unless-stopped
102+    deploy:
103+      resources:
104+        limits:
105+          cpus: '0.25'
102106     environment:
103107       TAP_DATABASE_URL: postgres://cospan:${POSTGRES_PASSWORD}@db:5432/tap
104108       TAP_BIND: ":2480"
@@ -127,6 +131,10 @@ services:
127131     image: ghcr.io/bluesky-social/indigo/tap:latest
128132     container_name: cospan-tap-knots
129133     restart: unless-stopped
134+    deploy:
135+      resources:
136+        limits:
137+          cpus: '0.25'
130138     environment:
131139       TAP_DATABASE_URL: postgres://cospan:${POSTGRES_PASSWORD}@db:5432/tap_knots
132140       TAP_BIND: ":2480"
cospan · schematic version control on atproto built on AT Protocol