Operator decision 2026-10-04 (sebastian): consensus-layer-synced op-stack nodes used to run
with --nodiscover / --disable-discovery and no published p2p port. A 0-peer EL can never
backfill a dropped-gossip gap, so op-node wedges in FCU-SYNCING until L1 derivation
catches up (katana 09-19..26, zircuit, mode). Blocks still arrive only via op-node; EL
peers are for gap-fill. Our own copies now discover each other implicitly.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RzxemENdcZBCxdGrzXiFvS
The tag ghcr.io/base/node:v1.1.1 does NOT exist (registry query returned
manifest unknown / HTTP 404 MANIFEST_UNKNOWN). With the repository
default now pointing at ghcr.io/base/node while the version pin still says
v1.1.1, base-sepolia currently resolves to an UNPULLABLE image reference,
breaking DEPLOY. v1.4.2 is the proven production image on base-mainnet
(carries Cobalt support, executed block 52000927) and resolves at
sha256:40e312a3f3aeb1b1599ad8d871aed20890cd0cd704be8633508694bad4f76abd.
Generated by Mistral Vibe.
Co-Authored-By: Mistral Vibe <vibe@mistral.ai>
- Upgrade base mainnet/sepolia reth and node versions from v1.1.0/v1.1.1/v1.2.0 to v1.4.2
- Regeneration includes recent main branch changes:
- polygon-mainnet bor: cap cache.gomemlimit to 12GB to prevent OOM
- gravity: add AptosBFT+Gravity-Reth execution node configs
- reference-rpc-endpoint.json: add mocachain URLs
Generated by Mistral Vibe.
Co-Authored-By: Mistral Vibe <vibe@mistral.ai>
DNS-only seeds burst-die (64 peers, 2s cumulative handshake timeout) and TBC
makes no dials at all; the header restore then stalls (968254 on 2026-09-29,
967460 on 2026-09-22) and op-node wedges on 'failed to fetch previous keystone'
because canonical BTC headers are missing from full TBC.
Pinned list matches the proven host-local workaround on us-16, where an nft
redirect (192.168.0.0/26:8333 -> :18333) serves /root/hemi-tbc-header-peer.py,
so these addresses only need to be dialable, not real bitcoin nodes. Landing
them in the tracked compose because regens wipe the .env COMPOSE_FILE append
(observed 09-29 03:06 and 03:56 UTC).
PR #5204 incorrectly bumped op.clients.reth.client_version (op-reth image)
from v2.4.3 to v2.5.2. The us-docker.pkg.dev/oplabs-tools-artifacts/images/op-reth
registry has no v2.5.2 tag, causing deploy failures on chains like bob.sepolia
and soneium.mainnet that inherit the default op-reth client without override.
This reverts only the op-reth version change, keeping the intended top-level
reth (ghcr.io/paradigmxyz/reth) bump to v2.5.2 intact.
Generated by Mistral Vibe.
Co-Authored-By: Mistral Vibe <vibe@mistral.ai>
claude
robandClaude Fable 5.1