zzstoatzz.io / relay
declarative relay deployment on hetzner relay-eval.waow.tech
atproto relay
14
fork

Configure Feed

Select the types of activity you want to include in your feed.

ops: right-size relay cluster resources

- jetstream: use internal service URL (ws://relay:2470) instead of
public ingress (wss://relay.waow.tech) — fixes websocket handshake
failures causing 30+ crash-loop restarts
- relay: lower memory limit 8Gi → 5Gi (actual ~4.2 GiB w/ 500K cache)
- collectiondir: bump memory limit 2Gi → 2560Mi (pebble index at
1.45 GiB and growing)
- reconnect cronjob: reduce job history 3 → 1

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

zzstoatzz 931bfbec 7a6ced5d

+5 -5
+1 -1
deploy/collectiondir-values.yaml
··· 42 42 memory: 128Mi 43 43 cpu: 50m 44 44 limits: 45 - memory: 2Gi 45 + memory: 2560Mi 46 46 47 47 defaultPodOptions: 48 48 imagePullSecrets:
+1 -1
deploy/jetstream-values.yaml
··· 10 10 repository: ghcr.io/bluesky-social/jetstream 11 11 tag: sha-05c89f0 12 12 env: 13 - JETSTREAM_WS_URL: "wss://relay.waow.tech/xrpc/com.atproto.sync.subscribeRepos" 13 + JETSTREAM_WS_URL: "ws://relay:2470/xrpc/com.atproto.sync.subscribeRepos" 14 14 JETSTREAM_LISTEN_ADDR: ":6008" 15 15 JETSTREAM_METRICS_LISTEN_ADDR: ":6009" 16 16 JETSTREAM_DATA_DIR: "/data"
+2 -2
deploy/reconnect-cronjob.yaml
··· 6 6 spec: 7 7 schedule: "0 */4 * * *" # every 4 hours 8 8 concurrencyPolicy: Forbid 9 - successfulJobsHistoryLimit: 3 10 - failedJobsHistoryLimit: 3 9 + successfulJobsHistoryLimit: 1 10 + failedJobsHistoryLimit: 1 11 11 jobTemplate: 12 12 spec: 13 13 backoffLimit: 1
+1 -1
deploy/relay-values.yaml
··· 36 36 memory: 1Gi 37 37 cpu: 500m 38 38 limits: 39 - memory: 8Gi 39 + memory: 5Gi 40 40 41 41 defaultPodOptions: 42 42 # hostNetwork recommended for full-network relays (high packet volume).