Enum VerifyErrorClass 

pub enum VerifyErrorClass {
    MissingComponent,
    DigestMismatch,
    CompressionInfoCorrupt,
    ChunkOffsetOutOfBounds,
    ChunkDecompressionError,
    UnsupportedCompressionFeature,
    UncompressedChunkCrcMismatch,
    UnexpectedEof,
    IndexEntryCorrupt,
    StatisticsHeaderCorrupt,
    SummaryCorrupt,
    BtiRootPointerCorrupt,
    BtiTrieCorrupt,
    RowScanFailed,
    OutOfOrderKeyOrRow,
    InvalidLocalDeletionTime,
    FilterFalseNegative,
}

Stable classification of a verification failure.

The variant is the machine-checkable “error code”; the VerifyFinding carries the human-readable context. These names are part of the verifier contract — callers (and CI) may match on them, so they must remain stable.

Variants

MissingComponent

A TOC.txt-listed component (or a structurally-required component) is absent from disk.

DigestMismatch

Digest.crc32 does not match the computed CRC32 of Data.db.

CompressionInfoCorrupt

CompressionInfo.db failed to parse, named an unsupported algorithm, or otherwise malformed (#1001).

ChunkOffsetOutOfBounds

A CompressionInfo.db chunk offset points outside Data.db.

ChunkDecompressionError

An inline Data.db chunk CRC32 did not match, or a chunk could not be read / decompressed (truncation, bit flip).

UnsupportedCompressionFeature

A chunk is compressed with a valid but UNSUPPORTED compression feature — distinct from truncation/bit-flip (ChunkDecompressionError) and from a checksum mismatch (DigestMismatch) (issue #1414). The canonical case is a zstd dictionary-compressed chunk: the frame is well-formed and its inline chunk CRC is valid, but CQLite ships no-dictionary zstd only, so the frame cannot be decoded. The reader fails closed with Error::UnsupportedFormat naming the feature (e.g. the Dictionary_ID); this class makes the verify report say “unsupported feature”, never “corruption”.

UncompressedChunkCrcMismatch

An uncompressed BIG Data.db chunk did not match its stored CRC.db per-chunk CRC32 (issue #1396) — the uncompressed analogue of the compressed path’s inline chunk-CRC finding (ChunkDecompressionError). Cassandra writes a CRC.db for every uncompressed BIG SSTable and verifies reads against it; a bit flip inside an uncompressed chunk is detected here (and, default-on, on every read). Also covers a truncated / short CRC.db (fewer per-chunk CRC entries than the Data.db has chunks). Reported via a VerifyFinding naming the failing chunk and the CRC.db/Data.db component.

UnexpectedEof

A component was truncated and a required read hit end-of-file.

IndexEntryCorrupt

Index.db (BIG) is structurally corrupt.

StatisticsHeaderCorrupt

Statistics.db header / body is corrupt.

SummaryCorrupt

Summary.db is truncated / unreadable.

BtiRootPointerCorrupt

BTI Partitions.db root pointer / node is corrupt.

BtiTrieCorrupt

BTI Rows.db trie is truncated / corrupt.

RowScanFailed

A full row scan failed for a reason not otherwise classified above.

OutOfOrderKeyOrRow

Partition keys are not in ascending on-disk (Murmur3 token) order, or clustering rows within a partition are not in ascending clustering order (issue #1282). Cassandra requires strictly ordered keys/rows; its sstableverify (SSTableIdentityIterator / Verifier) rejects an out-of-order key or row as corrupt.

InvalidLocalDeletionTime

A partition-level localDeletionTime is negative (invalid) on the legacy signed (nb) DeletionTime form (issue #1282). localDeletionTime is seconds since the Unix epoch; the only non-negative “special” value is the live sentinel i32::MAX (0x7FFFFFFF). A negative value cannot be a valid deletion time — Cassandra’s DeletionTime/Verifier treats it as corrupt. (The unsigned oa/da form legitimately represents far-future times in [2^31, 2^32), so those are NOT flagged — the on-disk format, not a heuristic, decides.)

FilterFalseNegative

A parseable BIG Filter.db reports “not present” (might_contain == false) for a partition key that IS present in the SSTable (its raw key bytes are enumerated from the authoritative Index.db) — a Bloom-filter FALSE NEGATIVE (issue #1398). Cassandra’s Filter.db carries no checksum, so a bit flipped from 1→0 inside the bit array is not detected on load and makes a live partition silently invisible on the BIG point-lookup path (partition_lookup.rs returns Ok(None) when the bloom says “miss”). Full scans and BTI (da) lookups are UNAFFECTED (they never gate on this bloom), so this is a detection tool Cassandra’s sstableverify lacks — Cassandra does not verify Filter.db contents and would report the same fixture clean.

Implementations

impl VerifyErrorClass

pub fn code(self) -> &'static str

Stable string code for the error class (used in reports / CI artifacts).

Trait Implementations

impl Clone for VerifyErrorClass

fn clone(&self) -> VerifyErrorClass

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for VerifyErrorClass

impl Debug for VerifyErrorClass

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Display for VerifyErrorClass

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for VerifyErrorClass

impl Hash for VerifyErrorClass

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for VerifyErrorClass

fn eq(&self, other: &VerifyErrorClass) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for VerifyErrorClass