apache/cassandra · error · InvalidRequestException

Could not parse TokenKey " + vs[i]

Error message

Could not parse TokenKey " + vs[i]

What it means

When parsing a routing-keys/token-list string from an Accord debug table, each comma-separated token that does not end the range syntax ('...]' style) is parsed as a TokenKey with the cluster partitioner. Any parse failure is rethrown as an InvalidRequestException identifying the specific token that could not be parsed.

Source

Thrown at src/java/org/apache/cassandra/db/virtual/AccordDebugKeyspace.java:2572

    }

    public static Participants<?> parseParticipants(Object input)
    {
        if (input == null)
            return null;

        String str = (String) input;
        if (str.isEmpty())
            return RoutingKeys.EMPTY;

        String[] vs = str.split("\\|");
        if (!vs[0].endsWith("]"))
        {
            RoutingKey[] keys = new RoutingKey[vs.length];
            for (int i = 0 ; i < keys.length ; ++i)
            {
                try { keys[i] = TokenKey.parse(vs[i], DatabaseDescriptor.getPartitioner()); }
                catch (Throwable t) { throw new InvalidRequestException("Could not parse TokenKey " + vs[i]); }
            }
            return RoutingKeys.of(keys);
        }
        else
        {
            TokenRange[] ranges = new TokenRange[vs.length];
            for (int i = 0 ; i < ranges.length ; ++i)
            {
                try { ranges[i] = TokenRange.parse(vs[i], DatabaseDescriptor.getPartitioner()); }
                catch (Throwable t) { throw new InvalidRequestException("Could not parse TokenKey " + vs[i]); }
            }
            return Ranges.of(ranges);
        }
    }

    public static String toString(Participants<?> participants)
    {
        StringBuilder out = new StringBuilder();

View on GitHub (pinned to 88fd0f6a0e)

Solutions

  1. Fix the offending token to be a valid literal for DatabaseDescriptor's partitioner (e.g. a plain signed long for Murmur3Token)
  2. Verify the whole string uses one consistent syntax: bare tokens for RoutingKeys, or full '[...]' range syntax for TokenRange
  3. Copy token keys directly from system tables (e.g. system.local or ring output) instead of hand-writing them

Example fix

// before
UPDATE system.accord_debug SET token_keys = '-9223372036854775808, 12a3';
// after
UPDATE system.accord_debug SET token_keys = '-9223372036854775808, 123'; // each token must parse with the cluster partitioner
Defensive patterns

Strategy: validation

Validate before calling

// each bare component must parse with the cluster partitioner before sending
for (String part : value.split(",")) {
    if (!part.trim().endsWith("]")) {
        try { TokenKey.parse(part.trim(), DatabaseDescriptor.getPartitioner()); }
        catch (Throwable t) { throw new IllegalArgumentException("bad TokenKey: " + part); }
    }
}

Try / catch

try {
    session.execute(update);
} catch (InvalidRequestException e) {
    if (e.getMessage().startsWith("Could not parse TokenKey")) {
        // fix the specific token echoed in the message and retry
    }
}

Prevention

When it happens

Trigger: Supplying a token key list column value where one of the individual token components (vs[i]) is not a valid token literal for the configured partitioner — e.g. malformed numeric token, stray characters, or a token from a different partitioner format.

Common situations: Hand-editing token ranges copied from another cluster with a different partitioner (Murmur3 vs Random vs ByteOrdered); truncating a copied list mid-token; mixing range syntax '[..]' with bare tokens incorrectly so a range fragment lands in the bare-token branch.

Understand the failure class

Background: "Invalid ... format", "must be in format X", "does not look like a ..." — invalid argument format errors across CLI tools and libraries — this error's family across 17 libraries.

Related errors

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/fdbebaf4ec3971cd. Report an issue: GitHub.