sstablesplit

Splits SSTable files into multiple SSTables of a maximum designated size while offline.

For example, if SizeTieredCompactionStrategy was used for a major compaction and results in an excessively large SSTable, split the table to ensure that compaction occurs before the next huge compaction.

Restriction: Stop HCD before you run this command.

The default location of this SSTable tool depends on the type of installation:

Package installations: /usr/bin/
Tarball installations: <installation_location>/resources/cassandra/tools/bin

Synopsis

sstablessplit
[--debug] [-h] [--no_snapshot]
[-s <max_size_in_MB>]
<sstable_filepath> [<sstable_filepath> ...]

Definition

The short form and long form parameters are comma-separated.

Command arguments

--debug: Display stack traces.
-h, --help: Display the usage and listing of the commands.
--no-snapshot: Do not snapshot SSTables before splitting.
-s, --size max_size_in_MB: Maximum size in MB for output SSTables. Default: 50.
sstable_filepath: Filepath to an SSTable.

Examples

Verify HCD is not running

nodetool status

Datacenter: Graph
================================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address        Load       Tokens       Owns    Host ID                               Rack
UN  10.200.177.92  265.04 KiB  1            ?       980cab6a-2e5d-44c6-b897-0733dde580ac  rack1
DN  10.200.177.94  426.21 KiB  1            ?       7ecbbc0c-627d-403e-b8cc-a2daa93d9ad3  rack1

Restriction: Stop HCD before you run this command.

Split SSTables to 10 MB

sstablesplit /var/lib/cassandra/data/cycling/cyclist_category-e1f76e21ce4311e8949e33016bf887c0/aa-1-bti-Statistics.db 10

Skipping inexisting file 10
        Skipping /var/lib/cassandra/data/cycling/cyclist_category-e1f76e21ce4311e8949e33016bf887c0/aa-1-bti-Data.db: it's size (0.000 MB) is less than the split size (50 MB)
        No sstables needed splitting.

Some tools require a larger heap size. For example, OutOfMemoryError exceptions may be resolved by increasing the heap size for sstablesplit.

To increase the heap size for sstablesplit to 8GB, change the following line in the tools/bin/sstablesplit shell script from:

MAX_HEAP_SIZE="256M"

to:

MAX_HEAP_SIZE="8G"