Conversation
DBDirName and ParseDBDirName make the shard-0x{full_shard_id} naming
convention a single point of truth: boot builds paths through it and the
inspect subcommand parses directory listings back through it.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
slave inspect --datadir <d> is read-only and config-free: it scans for
shard-0x{full_shard_id}/ chaindb directories, opens each in pebble
read-only mode, and prints the stored genesis metadata record and chain
head. A shard that cannot be opened or read is reported inline without
aborting the remaining shards, with a non-zero exit if any failed. An
absent metadata record is reported as an interrupted bootstrap rather
than an error, since the next boot re-runs the fresh path.
The README gains the inspect section and a fixtures section covering
provenance and the pyquarkchain cross-validation command.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
goleak now wraps both boot/shutdown smoke-test packages. The ignore list is empty on purpose: with metrics disabled neither geth nor pebble leaves a background goroutine behind after Stop(), and keeping it empty means the real chain's background work is heard here the day it lands. A new subprocess test initializes a datadir from the mainnet config and reruns the slave against it with the devnet config, asserting the run exits 1 and names the stored genesis, the config-derived genesis, the db path, and 'cluster config changed since initialization'. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…mmand Running the pyquarkchain cross-validation snippet with a bare system python fails at import time (no aiohttp); it needs a virtualenv with pyquarkchain's requirements installed. Both READMEs carrying the command now say so. Verified against both networks: the venv run reproduces the pinned mainnet 4036783e... and devnet 5ad443ef... hashes. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
syntrust
changed the title
syntrust
marked this pull request as ready for review
July 15, 2026 11:41
syntrust
requested review from
blockchaindevsh,
iteyelmp,
ping-ke,
qizhou and
qzhodl
July 15, 2026 11:41
slave-m3 now carries the final fixture helpers and goleak TestMain in run_test.go, so this branch's edits to slave_test.go and run_test.go dissolve into the merge; TestRunGenesisMismatchExitsLoudly relocates into inspect_test.go next to the initDataDir helper it depends on. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The README sections this branch reshuffled now originate upstream; its diff reduces to the 'Inspecting a datadir' and follow-up checklist sections. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The inspect labels move with it (meta version -> record version, genesis metadata -> genesis record), keeping the value column aligned. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
slave-m3 was merge-forwarded onto the base that #44 and #26 rewrote, so this merge takes its tree wholesale and re-applies only what M4 owns, ported to it. The GenesisRecord system this milestone was written against no longer exists. - cmd/slave/inspectcmd.go is rewritten around shard.ReadGenesisBlock: it prints the real block (hash, height, state root, coinbase and its amount map, both gas limits, hash_prev_root_block, x-shard cursor) instead of a record's fields. The misplaced-chaindb check now comes from the block itself, comparing its branch's full shard id against the directory name. - The inspect assertions are pinned against qkc/testdata/minor_genesis_golden.json, so what the report prints is pyquarkchain's own create_minor_block() output rather than a value derived a second way, and a new test covers the branch-versus-directory check. - The README's inspect docs move under Subcommands next to config and genesis, with the current output, and the follow-up checklist is rewritten: no record or fingerprint to retire, the scaffolding is the single genesis block key. Five old-world files this branch still carried are dropped rather than merged: qkc/genesis/, qkc/shard/genesis.go and their tests, and qkc/types/tokenbalances.go. The four commits superseded upstream were already dropped on slave-m3. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
inspect deserialized the stored genesis block and printed its fields, checking only that the block's branch matched the directory. A minor block's hash is its header's hash alone, so that left the fields hanging off the header unchecked: a database whose meta had been replaced still reported the original, authentic-looking block hash next to the substituted state root, and the command exited 0 with "0 failed". The boot path catches this by comparing the stored encoding against the config-derived block (shard.ReconcileGenesisBlock), but inspect is config-free and has no such comparison to fall back on. Rehash the meta and check it against the hash_meta the header commits to, which is the check that closes the gap, and require block 0 to be block 0 with an empty body and both sub-structs present. None of these needs the cluster config. Move every check, including the misplaced-chaindb one, ahead of all printing, so a shard that fails is described by its error alone. Printing first and reporting after put a state root nobody should trust under a hash that looked genuine, and contradicted what the README already claimed. Stop reading "no genesis block" as an interrupted bootstrap unconditionally. The block is written last, once the chain stands, so its absence means an interrupted bootstrap only on a database holding nothing else; a head pointer with no genesis under it is unreachable for this lifecycle, and may not be a shard chaindb at all. Report it instead of claiming the next boot re-initializes it safely. Read the head hash through Has/Get rather than rawdb.ReadHeadBlockHash, which discards its Get error and answers a failed read with the zero hash — printed as "none recorded", a claim about a database that could not be read. Latch report write errors in a sticky writer and fail on them, so a report that never reached the reader is not summarized as a success. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
An initialized shard chaindb holds three things: the genesis block, the state it commits to, and the EVM rule set. inspect reported the first two and said nothing about the third, although it is the other half of what a reopen is checked against — a shard can stand on exactly the right genesis and still refuse to boot on an incompatible fork schedule, and nothing showed what schedule was stored. Print the chain id and the fork schedule alongside the block. The schedule is rendered from the stored encoding rather than a fixed list of fork fields, so a rule set carrying a fork this build does not know about is shown rather than silently dropped, and it is ordered by activation, which is how a schedule reads. A datadir initialized before its rule set was written reports that instead of looking complete. It is not a failure: ReconcileChainConfig treats a missing rule set as recoverable and answers it by warning and writing one, and inspect follows that judgement rather than inventing a stricter one. Read the key directly instead of through rawdb.ReadChainConfig, which answers a failed read and a malformed encoding alike with a nil config — indistinguishable here from a rule set that was never stored, the same way ReadHeadBlockHash cannot tell an unreadable head from an absent one. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
M4 of #17, stacked on #38 (base:
slave-m3). Merge-forwarded onto the rewrittenslave-m3(post-#44/#26).What
Adds
slave inspect --datadir <d>: a read-only, config-free look at what a slave's datadir actually holds.Test
create_minor_block()golden values, not against anything re-derived here.-race, gofmt and vet clean.