apache/cassandra · error · ArithmeticException

Natural log of number zero or less

Error message

Natural log of number zero or less

What it means

DecimalType.log computes the natural logarithm of a decimal operand; since ln is undefined for zero and negative numbers, it raises ArithmeticException('Natural log of number zero or less') when the input compares <= 0.

Solutions

1. Filter rows in the query: WHERE col > 0 before applying ln()
2. Use a CASE/UDF to map non-positive values to null instead of ln(col)
3. Clean the data (remove/fix non-positive values) if they are invalid

Example fix

// before
SELECT ln(delta) FROM t;
// after
SELECT ln(delta) FROM t WHERE delta > 0;

if (input == null || input.signum() <= 0) throw new IllegalArgumentException("ln requires a strictly positive value");

boolean isPositive(BigDecimal v) { return v != null && v.signum() > 0; }

try { result = DecimalType.instance.log(value); } catch (ArithmeticException e) { /* map non-positive input to null/default */ }

When it happens

Trigger: Executing a CQL expression like ln(col) or the function-call path that reaches log(Number) with a value of 0 or a negative decimal.

Common situations: Queries over columns that legitimately hold zeros/negatives (deltas, balances) without filtering; bad data ingested into measured quantities expected positive.

Understand the failure class

Background: "value must be between 0 and 1" / "out of range" / "must not be negative" errors: fixing range-validation failures across open-source libraries — this error's family across 42 libraries.

Related errors

• Log10 of number zero or less
• A NaN cannot be converted into a decimal
• An infinite number cannot be converted into a decimal
• Cannot parse decimal value from
• Expected 0 or at least 4 bytes

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/538c281b62a88d9a. Report an issue: GitHub.

Appendix: source

Thrown at src/java/org/apache/cassandra/db/marshal/DecimalType.java:438

    }

    protected BigDecimal exp(BigDecimal input)
    {
        int precision = input.precision();
        precision = Math.max(MIN_SIGNIFICANT_DIGITS, precision);
        precision = Math.min(MAX_PRECISION.getPrecision(), precision);
        return BigDecimalMath.exp(input, new MathContext(precision, RoundingMode.HALF_EVEN));
    }

    @Override
    public ByteBuffer log(Number input)
    {
        return decompose(log(toBigDecimal(input)));
    }

    protected BigDecimal log(BigDecimal input)
    {
        if (input.compareTo(BigDecimal.ZERO) <= 0) throw new ArithmeticException("Natural log of number zero or less");
        int precision = input.precision();
        precision = Math.max(MIN_SIGNIFICANT_DIGITS, precision);
        precision = Math.min(MAX_PRECISION.getPrecision(), precision);
        return BigDecimalMath.log(input, new MathContext(precision, RoundingMode.HALF_EVEN));
    }

    @Override
    public ByteBuffer log10(Number input)
    {
        return decompose(log10(toBigDecimal(input)));
    }

    protected BigDecimal log10(BigDecimal input)
    {
        if (input.compareTo(BigDecimal.ZERO) <= 0) throw new ArithmeticException("Log10 of number zero or less");
        int precision = input.precision();
        precision = Math.max(MIN_SIGNIFICANT_DIGITS, precision);
        precision = Math.min(MAX_PRECISION.getPrecision(), precision);

View on GitHub (pinned to 88fd0f6a0e)