apache/cassandra · critical · FSWriteError

No configured data directory contains enough space to write

Error message

No configured data directory contains enough space to write " + writeSize + " bytes

What it means

Directories.getWriteableLocationAsFile(writeSize) picks a data directory that can hold a write of the given size. If no configured data directory passes the disk-usage policy (has enough free space after writeSize, not blacklisted, not read-only), it throws FSWriteError wrapping an IOException stating no directory contains enough space.

Source

Thrown at src/java/org/apache/cassandra/db/Directories.java:367

     * Do not use this method in production code.
     *
     * @throws FSWriteError if all directories are disallowed.
     */
    public File getDirectoryForNewSSTables()
    {
        return getWriteableLocationAsFile(-1L);
    }

    /**
     * Returns an allowed directory that _currently_ has {@code writeSize} bytes as usable space.
     *
     * @throws FSWriteError if all directories are disallowed.
     */
    public File getWriteableLocationAsFile(long writeSize)
    {
        File location = getLocationForDisk(getWriteableLocation(writeSize));
        if (location == null)
            throw new FSWriteError(new IOException("No configured data directory contains enough space to write " + writeSize + " bytes"), "");
        return location;
    }

    /**
     * Returns a data directory to load the file {@code sourceFile}. If the sourceFile is on same disk partition as any
     * data directory then use that one as data directory otherwise use {@link #getWriteableLocationAsFile(long)} to
     * find suitable data directory.
     *
     * Also makes sure returned directory is not disallowed.
     *
     * @throws FSWriteError if all directories are disallowed.
     */
    public File getWriteableLocationToLoadFile(final File sourceFile)
    {
        try
        {
            final FileStore srcFileStore = Files.getFileStore(sourceFile.toPath());
            for (final File dataPath : dataPaths)

View on GitHub (pinned to 88fd0f6a0e)

Solutions

1. Free space on the data directories (delete snapshots via `nodetool clearsnapshot`, run compaction to reclaim space) or add/expand disks.
2. Move data_file_directories to larger volumes and restart, or rebalance tokens/tables to nodes with space.
3. Lower disk_usage_percent settings only as a stopgap; verify with `nodetool diskusage` / `df -h`.
4. Reduce flush/compaction write sizes (smaller compaction strategies, e.g. LCS/STCS tuning) if a single operation is larger than any disk.

Example fix

// cassandra.yaml before
data_file_directories: [/var/lib/cassandra/data]
// after: distribute across larger volumes
data_file_directories:
  - /mnt/disk1/cassandra/data
  - /mnt/disk2/cassandra/data

long usable = file.getTotalSpace() - file.getUsableSpace();
if (writeSize > file.getUsableSpace() * (1 - 0.10)) // keep 10% headroom
    throw new IllegalStateException("Insufficient disk space for " + writeSize);

try {
    File dir = directories.getWriteableLocationAsFile(writeSize);
} catch (FSWriteError e) {
    if (e.getMessage().contains("No configured data directory contains enough space"))
        clearSnapshotsAndAlert(); // nodetool clearsnapshot, page on-call
    throw e;
}

When it happens

Trigger: Flushing a memtable, compacting, or writing an SSTable whose estimated size exceeds the free space of every configured data_file_directories location, taking into account disk_usage_percent_warn/fail thresholds and read-only/blacklisted disks.

Common situations: Data directories actually full or nearly full (small disk, unbounded growth); all data dirs on one small partition; incorrect data_file_directories config pointing to tiny mounts; disk failures leaving only undersized volumes; large compactions or huge flushes after big writes.

Understand the failure class

Background: "failed to write file", "Could not save figure", "Error saving remote file" — file write failed: causes and fixes across languages and libraries — this error's family across 38 libraries.

Related errors

• Unsupported disk access mode for background_write_disk_acces
• Not enough space for compaction (%s) of %s.%s, estimated sst
• Not enough space to write %s to %s (%s available)
• Not enough disk space to store %s
• Only {} free across all data volumes. Consider adding more c

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/bff7d7bf554b6204. Report an issue: GitHub.