Fork-time fixes: doma, hashkeychain, katana-testnet (riding reth v2.5.2 bump) #86

Merged
sebastian merged 2 commits from vupd/5202-reth-v2.5.2 into main 2026-09-27 06:15:27 +00:00
Collaborator

Summary

General fork-time audit (check_fork_times.py --json) run alongside the reth v2.4.1→v2.5.2 bump (vibe-node #5202) found PAST activation gaps — the live sequencer's optimism_rollupConfig reports fork times not yet reflected in our bind-mounted rollup.json/genesis.json. These nodes are already forked or about to be; fixed here via check_fork_times.py --fix --only <chain>/<network>.

Unrelated to the reth client itself — this is the standing per-bump fork-time sweep.

Fixed (PAST — node is/was already forked)

Chain/network Fork New activation Sequencer
doma/mainnet karst_time 1787673601 https://rpc.doma.xyz
doma/testnet karst_time 1787068801 https://rpc-testnet.doma.xyz
hashkeychain/mainnet holocene_time 1785306600 published (may lag)
hashkeychain/mainnet isthmus_time 1785306900 published (may lag)
hashkeychain/mainnet jovian_time 1785307200 published (may lag)
katana/testnet isthmus_time 1749807000 https://rpc-bokuto.katanarpc.com
katana/testnet jovian_time 1770136201 https://rpc-bokuto.katanarpc.com
katana/testnet karst_time 1787760001 https://rpc-bokuto.katanarpc.com

Listed only — UNVERIFIED (sequencer unreachable, not guessed)

  • hemi/mainnet, hemi/sepolia
  • manta-pacific/mainnet
  • xlayer/mainnet, xlayer/testnet

Deploy note

These rollup.json/genesis.json files are bind-mounted — a plain deploy-compose-repo regen will NOT restart the affected containers. After merge+deploy, run force-recreate <host> <node_path> for each of doma-mainnet, doma-testnet, hashkeychain-mainnet, and katana-testnet, then confirm the new fork name appears in the op-node/op-reth logs.

Since these chains were already past their fork activation when this landed, treat each as already forked: this is a config-format correction to match reality, not a preventive fix — confirm chain health after force-recreate rather than assuming a clean restart.

Companion PR

Must merge together with vibe-node #5202 (reth v2.4.1→v2.5.2 context.yml bump, same branch name vupd/5202-reth-v2.5.2).

🤖 Generated with Claude Code

## Summary General fork-time audit (`check_fork_times.py --json`) run alongside the reth v2.4.1→v2.5.2 bump (vibe-node #5202) found PAST activation gaps — the live sequencer's `optimism_rollupConfig` reports fork times not yet reflected in our bind-mounted `rollup.json`/`genesis.json`. These nodes are already forked or about to be; fixed here via `check_fork_times.py --fix --only <chain>/<network>`. Unrelated to the reth client itself — this is the standing per-bump fork-time sweep. ## Fixed (PAST — node is/was already forked) | Chain/network | Fork | New activation | Sequencer | |---|---|---|---| | doma/mainnet | karst_time | 1787673601 | https://rpc.doma.xyz | | doma/testnet | karst_time | 1787068801 | https://rpc-testnet.doma.xyz | | hashkeychain/mainnet | holocene_time | 1785306600 | published (may lag) | | hashkeychain/mainnet | isthmus_time | 1785306900 | published (may lag) | | hashkeychain/mainnet | jovian_time | 1785307200 | published (may lag) | | katana/testnet | isthmus_time | 1749807000 | https://rpc-bokuto.katanarpc.com | | katana/testnet | jovian_time | 1770136201 | https://rpc-bokuto.katanarpc.com | | katana/testnet | karst_time | 1787760001 | https://rpc-bokuto.katanarpc.com | ## Listed only — UNVERIFIED (sequencer unreachable, not guessed) - hemi/mainnet, hemi/sepolia - manta-pacific/mainnet - xlayer/mainnet, xlayer/testnet ## Deploy note These `rollup.json`/`genesis.json` files are **bind-mounted** — a plain `deploy-compose-repo` regen will NOT restart the affected containers. After merge+deploy, run `force-recreate <host> <node_path>` for each of doma-mainnet, doma-testnet, hashkeychain-mainnet, and katana-testnet, then confirm the new fork name appears in the op-node/op-reth logs. Since these chains were already past their fork activation when this landed, treat each as **already forked**: this is a config-format correction to match reality, not a preventive fix — confirm chain health after force-recreate rather than assuming a clean restart. ## Companion PR Must merge together with vibe-node #5202 (reth v2.4.1→v2.5.2 context.yml bump, same branch name `vupd/5202-reth-v2.5.2`). 🤖 Generated with [Claude Code](https://claude.com/claude-code)
claude added 2 commits 2026-09-24 10:04:12 +00:00
sebastian merged commit 9a130a7581 into main 2026-09-27 06:15:27 +00:00
sebastian deleted branch vupd/5202-reth-v2.5.2 2026-09-27 06:15:34 +00:00
Author
Collaborator

Deployed + verified.

Inventory check (list-inventory, 2026-09-27) found only hashkeychain/mainnet actually deployed (op/geth/hashkeychain-mainnet-op-geth-pruned-pebble-path on rpc-sg-9). doma/mainnet, doma/testnet, and katana/testnet are not deployed anywhere in the fleet (0 hosts) — dormant-safe: the corrected rollup.json/genesis.json is now in main and will apply automatically whenever/if those chains are ever deployed. (Note: katana-mainnet, running on 4 hosts, is a different chain unaffected by this PR's katana/testnet fix.)

For the one live chain:

  • deploy-compose-repo scoped to rpc-sg-9 (job 709942af) — regenerated + pushed rpc main (9a130a758123→7b2872b1564e), deploy OK.
  • force-recreate rpc-sg-9 op/geth/hashkeychain-mainnet-op-geth-pruned-pebble-path (job 4fb59f8c) — bind-mounted files needed the explicit recreate; both containers recreated + started cleanly.
  • Verified via run-show-status sg-9 + sync-status: hashkeychain-mainnet: online. (Note: rpc-sg-9 dockerd is scan-blind, docker logs refused — couldn't grep the exact fork-name line from container logs directly, but online status + clean recreate is the available health signal.)

Closing out the fork-times tracking claim for this PR.

**Deployed + verified.** Inventory check (list-inventory, 2026-09-27) found only **hashkeychain/mainnet** actually deployed (`op/geth/hashkeychain-mainnet-op-geth-pruned-pebble-path` on `rpc-sg-9`). `doma/mainnet`, `doma/testnet`, and `katana/testnet` are **not deployed anywhere in the fleet** (0 hosts) — dormant-safe: the corrected `rollup.json`/`genesis.json` is now in `main` and will apply automatically whenever/if those chains are ever deployed. (Note: `katana-mainnet`, running on 4 hosts, is a different chain unaffected by this PR's `katana/testnet` fix.) For the one live chain: - `deploy-compose-repo` scoped to `rpc-sg-9` (job `709942af`) — regenerated + pushed rpc main (`9a130a758123`→`7b2872b1564e`), deploy OK. - `force-recreate rpc-sg-9 op/geth/hashkeychain-mainnet-op-geth-pruned-pebble-path` (job `4fb59f8c`) — bind-mounted files needed the explicit recreate; both containers recreated + started cleanly. - Verified via `run-show-status sg-9` + `sync-status`: `hashkeychain-mainnet: online`. (Note: `rpc-sg-9` dockerd is scan-blind, `docker logs` refused — couldn't grep the exact fork-name line from container logs directly, but online status + clean recreate is the available health signal.) Closing out the fork-times tracking claim for this PR.
Sign in to join this conversation.
No Reviewers
No labels
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: StakeSquid/ethereum-rpc-docker#86