op-geth does not support admin_addStaticPeer; classify op-geth via
compose path (admin_nodeInfo often still says Geth) and fall back on
go-ethereum-style "does not exist/is not available" errors so reinject
and connect-peers restore peers (e.g. hashkey-mainnet).
ss1 shipped without drpcorg public-submodule resources and crash-loops
at boot (compatible-clients.yaml missing); ss2 is the boot-verified fix.
Co-authored-by: Cursor <cursoragent@cursor.com>
Fleet-wide default is now stakesquid/dshackle:0.79.11-ss1 (digest-pinned);
DSHACKLE_IMAGE env escape hatch retained for rollback to upstream.
Co-authored-by: Cursor <cursoragent@cursor.com>
Default remains drpcorg/dshackle:0.79.10; canary hosts can set
DSHACKLE_IMAGE via host_vars env_overrides without editing compose.
Co-authored-by: Cursor <cursoragent@cursor.com>
Orphaned op-geth files (already removed from config.yml) plus sibling
op-reth/op-erigon profiles still share Traefik service names/paths under
COMPOSE_FILE. Keep only ink-mainnet-op-reth-pruned-trace--kona.yml as the
single active /ink-mainnet route.
EL/CL baseFee mismatch at block 22961752; CL crash-loops on
INVALID payload from post-v1.4.2 block format while beacon-kit
is on v1.4.1. Also regenerates ink-mainnet OPE P2P bootnodes
from parent a790c3e and refreshes reference-rpc-endpoint (mova).
Co-authored-by: Cursor <cursoragent@cursor.com>
Image default is -Xms1024M with no -Xmx, so heap can grow unbounded (~30GiB).
us-41 was at 90% memory with no swap; -Xmx4g provides a safety limit.
Co-authored-by: Cursor <cursoragent@cursor.com>
Canonicalize plasma-testnet so the tracked compose adopts the same
volume/service names as the de-32 hand-written override (rpc_plasma-
testnet-reth-pruned-trace-{execution,consensus,jwt}). Also refresh
mainnet observer sed to be chain-agnostic.
Also regenerate compose for already-merged hashkey static-peers / op-geth
static-nodes entrypoint (parent #1437) that had not yet landed in rpc.
Co-authored-by: Cursor <cursoragent@cursor.com>
Replace buggy && short-circuit with if/fi so the one-shot init
exits 0 when the signing key is already present (matches mainnet/
moderato and templates/clients/reth.tempo.yml), unblocking sync
resume on data-intact hosts still on the orphaned testnet compose.
Insert validate-dshackle-1to1.py between the overrides weave and the HUP signal:
a dshackle config routing >1 node for the same chain is REFUSED (exit 1) so the
live config stays active. Structural enforcement of 1 dshackle = 1 node/chain —
we can't attribute traffic to multiple nodes behind one proxy (no per-upstream
request metric; conn-seconds biases it). See /root/proxy-1to1-invariant-plan.md.
Co-Authored-By: Claude <noreply@anthropic.com>
Refuses to activate a dshackle config routing >1 upstream for the same chain
(any kind/client) — we cannot attribute traffic to multiple nodes behind one
proxy (no per-upstream request-count metric; conn-seconds biases it), and the
attribution model + placement planner both assume 1:1. Wire into
reload_dshackle.sh BEFORE the reload so a drifting config can never go live.
Host-local by design (the repo doesn't track host->dshackle assignment; 1:1 is
only checkable against the assembled dshackle config at /root/rpc/main_configs).
See /root/proxy-1to1-invariant-plan.md.
Co-Authored-By: Claude <noreply@anthropic.com>
Empty ws-origins rejects WebSocket connections with an Origin header
(dshackle head-subscription monitoring). Sed under [json-rpc] like
address/ws-address so restarts keep the value.
Node was forked at block 30194654 because rollup.json lacked holocene_time
and later fork timestamps; add granite/holocene/isthmus/jovian times so
op-node can apply post-Holocene batches after resync.
Canyon-fork EIP-1559 denominator, present in the authoritative genesis
on rpc-de-32 but missing from the committed copy. Matches the fork-time
addition in d4fc9487.
Add the post-Regolith OP-stack fork activation timestamps (canyon through
karst) to op/mode/mainnet/genesis.json, sourced from the authoritative
copy running on rpc-de-32 (sha256 ea320f6a). The committed genesis was
missing every fork time after regolithTime, so op-reth could not apply
the L2 hardfork schedule.
canyonTime=shanghaiTime=1704992401, deltaTime=1708560000,
ecotoneTime=1710374401, fjordTime=1720627201, graniteTime=1726070401,
holoceneTime=1736445601, isthmusTime=1746806401, jovianTime=1764691201,
karstTime=1783526401.
$temp_file is only created when curl is invoked with -o (the -o loop
above), but the post-success null/error check always ran `cat "$temp_file"`.
When a caller omits -o (e.g. check-health.sh's Cosmos reference RPC), the
variable is empty, so bash executed `cat ""` and printed
"cat: '': No such file or directory" — once per show-status run.
In that no-o case the body is already captured in $output, so read the
response from $output when $temp_file is unset, and from $temp_file when
it exists. Behaviour with -o is unchanged; the result==null / JSON-RPC
error skip-to-next-URL fallback is preserved in both paths.
Co-Authored-By: Claude <noreply@anthropic.com>
Match ronin-mainnet's EigenDA alt-DA recipe on both saigon profiles
(archive-trace + pruned-trace): OP_NODE_SYNCMODE execution-layer ->
consensus-layer and OP_NODE_L2_SKIP_SYNC_START_CHECK false -> true.
Hand-maintained edit (these CL-sync flags are not template-driven);
do NOT regenerate saigon via update.sh or it reverts. Mainnet untouched.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Custom op-node forks on these chains cannot run kona-node; drop the
orphaned --kona.yml compose files and registry entries.
Co-authored-by: Cursor <cursoragent@cursor.com>
claude
