Influx Inspect disk utility

Influx Inspect is an InfluxDB disk utility that can be used to:

  • View detailed information about disk shards.
  • Export data from a shard to InfluxDB line protocol that can be inserted back into the database.
  • Convert TSM index shards to TSI index shards.

influx_inspect utility

Syntax

  1. influx_inspect [ [ command ] [ options ] ]

-help is the default command and prints syntax and usage information for the tool.

influx_inspect commands

The influx_inspect commands are summarized here, with links to detailed information on each of the commands.

  • buildtsi: Converts in-memory (TSM-based) shards to TSI.
  • deletetsm: Bulk deletes a measurement from a raw TSM file.
  • dumptsi: Dumps low-level details about TSI files.
  • dumptsm: Dumps low-level details about TSM files.
  • dumptsmwal: Dump all data from a WAL file.

  • export: Exports raw data from a shard in InfluxDB line protocol format.

  • report: Displays a shard level report.
  • reporttsi: Reports on cardinality for measurements and shards.
  • verify: Verifies the integrity of TSM files.
  • verify-seriesfile: Verifies the integrity of series files.
  • verify-tombstone: Verifies the integrity of tombstones.

buildtsi

Builds TSI (Time Series Index) disk-based shard index files and associated series files. The index is written to a temporary location until complete and then moved to a permanent location. If an error occurs, then this operation will fall back to the original in-memory index.

Note: For offline conversion only. When TSI is enabled, new shards use the TSI indexes. Existing shards continue as TSM-based shards until converted offline.

Syntax
  1. influx_inspect buildtsi -datadir <data_dir> -waldir <wal_dir> [ options ]

Note: Use the buildtsi command with the user account that you are going to run the database as, or ensure that the permissions match after running the command.

Options

Optional arguments are in brackets.

[ -batch-size ]

The size of the batches written to the index. Default value is 10000.

Warning: Setting this value can have adverse effects on performance and heap size.

[ -compact-series-file ]

Does not rebuild the index. Compacts the existing series file, including offline series. Iterates series in each segment and rewrites non-tombstoned series in the index to a new .tmp file next to the segment. Once all segments are converted, the temporary files overwrite the original segments.

[ -concurrency ]

The number of workers to dedicate to shard index building. Defaults to GOMAXPROCS value.

[ -database <db_name> ]

The name of the database.

-datadir <data_dir>

The path to the data directory.

[ -max-cache-size ]

The maximum size of the cache before it starts rejecting writes. This value overrides the configuration setting for [data] cache-max-memory-size. Default value is 1073741824.

[ -max-log-file-size ]

The maximum size of the log file. Default value is 1048576.

[ -retention <rp_name> ]

The name of the retention policy.

[ -shard <shard_ID> ]

The identifier of the shard.

[ -v ]

Flag to enable output in verbose mode.

-waldir <wal_dir>

The directory for the WAL (Write Ahead Log) files.

Examples

Converting all shards on a node
  1. $ influx_inspect buildtsi -datadir ~/.influxdb/data -waldir ~/.influxdb/wal
Converting all shards for a database
  1. $ influx_inspect buildtsi -database mydb -datadir ~/.influxdb/data -waldir ~/.influxdb/wal
Converting a specific shard
  1. $ influx_inspect buildtsi -database stress -shard 1 -datadir ~/.influxdb/data -waldir ~/.influxdb/wal

deletetsm

Use deletetsm -measurement to delete a measurement in a raw TSM file (from specified shards). Use deletetsm -sanitize to remove all tag and field keys containing non-printable Unicode characters in a raw TSM file (from specified shards).

Warning: Use the deletetsm command only when your InfluxDB instance is offline (influxd service is not running).

Syntax

  1. influx_inspect deletetsm -measurement <measurement_name> [ arguments ] <path>
<path>

Path to the .tsm file, located by default in the data directory.

When specifying the path, wildcards (*) can replace one or more characters.

Options

Either the -measurement or -sanitize flag is required.

-measurement

The name of the measurement to delete from TSM files.

-sanitize

Flag to remove all keys containing non-printable Unicode characters from TSM files.

-v

Optional. Flag to enable verbose logging.

Examples

Delete a measurement from a single shard

Delete the measurement h2o_feet from a single shard.

  1. ./influx_inspect deletetsm -measurement h2o_feet /influxdb/data/location/autogen/1384/*.tsm
Delete a measurement from all shards in the database

Delete the measurement h2o_feet from all shards in the database.

  1. ./influx_inspect deletetsm -measurement h2o_feet /influxdb/data/location/autogen/*/*.tsm

dumptsi

Dumps low-level details about TSI files, including .tsl log files and .tsi index files.

Syntax

  1. influx_inspect dumptsi [ options ] <index_path>

If no options are specified, summary statistics are provided for each file.

Options

Optional arguments are in brackets.

-series-file <series_path>

Path to the _series directory under the database data directory. Required.

[ -series ]

Dump raw series data.

[ -measurements ]

Dump raw measurement data.

[ -tag-keys ]

Dump raw tag keys.

[ -tag-values ]

Dump raw tag values.

[ -tag-value-series ]

Dump raw series for each tag value.

[ -measurement-filter <regular_expression> ]

Filter data by measurement regular expression.

[ -tag-key-filter <regular_expression> ]

Filter data by tag key regular expression.

[ -tag-value-filter <regular_expresssion> ]

Filter data by tag value regular expression.

Examples

Specifying paths to the _series and index directories
  1. $ influx_inspect dumptsi -series-file /path/to/db/_series /path/to/index
Specifying paths to the _series directory and an index file
  1. $ influx_inspect dumptsi -series-file /path/to/db/_series /path/to/index/file0
Specifying paths to the _series directory and multiple index files
  1. $ influx_inspect dumptsi -series-file /path/to/db/_series /path/to/index/file0 /path/to/index/file1 ...

dumptsm

Dumps low-level details about TSM files, including TSM (.tsm) files and WAL (.wal) files.

Syntax

  1. influx_inspect dumptsm [ options ] <path>
<path>

Path to the .tsm file, located by default in the data directory.

Options

Optional arguments are in brackets.

[ -index ]

Flag to dump raw index data. Default value is false.

[ -blocks ]

Flag to dump raw block data. Default value is false.

[ -all ]

Flag to dump all data. Caution: This may print a lot of information. Default value is false.

[ -filter-key <key_name> ]

Display only index data and block data that match this key substring. Default value is "".

dumptsmwal

Dumps all entries from one or more WAL (.wal) files only and excludes TSM (.tsm) files.

Syntax

  1. influx_inspect dumptsmwal [ options ] <wal_dir>

Options

Optional arguments are in brackets.

[ -show-duplicates ]

Flag to show keys which have duplicate or out-of-order timestamps. If a user writes points with timestamps set by the client, then multiple points with the same timestamp (or with time-descending timestamps) can be written.

export

Exports all TSM files in InfluxDB line protocol data format. Writes all WAL file data for _internal/monitor. This output file can be imported using the influx command.

Syntax

  1. influx_inspect export [ options ]

Options

Optional arguments are in brackets.

[ -compress ]

The flag to compress the output using gzip compression. Default value is false.

[ -database <db_name> ]

The name of the database to export. Default value is "".

-datadir <data_dir>

The path to the data directory. Default value is "$HOME/.influxdb/data".

[ -end <timestamp> ]

The timestamp for the end of the time range. Must be in RFC3339 format.

RFC3339 requires very specific formatting. For example, to indicate no time zone offset (UTC+0), you must include Z or +00:00 after seconds. Examples of valid RFC3339 formats include:

No offset

  1. YYYY-MM-DDTHH:MM:SS+00:00
  2. YYYY-MM-DDTHH:MM:SSZ
  3. YYYY-MM-DDTHH:MM:SS.nnnnnnZ (fractional seconds (.nnnnnn) are optional)

With offset

  1. YYYY-MM-DDTHH:MM:SS-08:00
  2. YYYY-MM-DDTHH:MM:SS+07:00

Note: With offsets, avoid replacing the + or - sign with a Z. It may cause an error or print Z (ISO 8601 behavior) instead of the time zone offset.

[ -lponly ]

Output data in line protocol format only. Does not include comments or data definition language (DDL), like CREATE DATABASE.

[ -out <export_dir> ]

The location for the export file. Default value is "$HOME/.influxdb/export".

[ -retention <rp_name> ]

The name of the retention policy to export. Default value is "".

[ -start <timestamp> ]

The timestamp for the start of the time range. The timestamp string must be in RFC3339 format.

[ -waldir <wal_dir> ]

Path to the WAL directory. Default value is "$HOME/.influxdb/wal".

Examples

Export all databases and compress the output
  1. influx_inspect export -compress
Export data from a specific database and retention policy
  1. influx_inspect export -database mydb -retention autogen
Output file
  1. # DDL
  2. CREATE DATABASE MY_DB_NAME
  3. CREATE RETENTION POLICY autogen ON MY_DB_NAME DURATION inf REPLICATION 1
  4. # DML
  5. # CONTEXT-DATABASE:MY_DB_NAME
  6. # CONTEXT-RETENTION-POLICY:autogen
  7. randset value=97.9296104805 1439856000000000000
  8. randset value=25.3849066842 1439856100000000000

report

Displays series metadata for all shards. The default location is $HOME/.influxdb.

Syntax

  1. influx_inspect report [ options ]

Options

Optional arguments are in brackets.

[ -pattern "<regular expression/wildcard>" ]

The regular expression or wildcard pattern to match included files. Default value is "".

[ -detailed ]

The flag to report detailed cardinality estimates. Default value is false.

[ -exact ]

The flag to report exact cardinality counts instead of estimates. Default value is false. Note: This can use a lot of memory.

reporttsi

The report does the following:

  • Calculates the total exact series cardinality in the database.
  • Segments that cardinality by measurement, and emits those cardinality values.
  • Emits total exact cardinality for each shard in the database.
  • Segments for each shard the exact cardinality for each measurement in the shard.
  • Optionally limits the results in each shard to the “top n”.

The reporttsi command is primarily useful when there has been a change in cardinality and it’s not clear which measurement is responsible for this change, and further, when that change happened. Estimating an accurate cardinality breakdown for each measurement and for each shard will help answer those questions.

Syntax

  1. influx_inspect reporttsi -db-path <path-to-db> [ options ]

Options

Optional arguments are in brackets.

-db-path <path-to-db>

The path to the database.

[ -top <n> ]

Limits the results to the top specified number within each shard.

Performance

The reporttsi command uses simple slice/maps to store low cardinality measurements, which saves on the cost of initializing bitmaps. For high cardinality measurements the tool uses roaring bitmaps, which means we don’t need to store all series IDs on the heap while running the tool. Conversion from low-cardinality to high-cardinality representations is done automatically while the tool runs.

verify

Verifies the integrity of TSM files.

Syntax

  1. influx_inspect verify [ options ]

Options

Optional arguments are in brackets.

-dir <storage_root>

The path to the storage root directory. ​Default value is "/root/.influxdb".

verify-seriesfile

Verifies the integrity of series files.

Syntax

  1. influx_inspect verify-seriesfile [ options ]

Options

Optional arguments are in brackets.

[ -c <number> ]

Specifies the number of concurrent workers to run for this command. Default is equal to the value of GOMAXPROCS. If performance is adversely impacted, you can set a lower value.

[ -dir <path> ]

Specifies the root data path. Defaults to ~/.influxdb/data.

[ -db <db_name> ]

Restricts verifying series files to the specified database in the data directory.

[ -series-file <path> ]

Path to a specific series file; overrides -db and -dir.

[ -v ]

Enables verbose logging.

verify-tombstone

Verifies the integrity of tombstones.

Syntax

  1. influx_inspect verify-tombstone [ options ]

Finds and verifies all tombstones under the specified directory path (by default, ~/.influxdb/data). Files are verified serially.

Options

Optional arguments are in brackets.

[ -dir <path> ]

Specifies the root data path. Defaults to ~/.influxdb/data. This path can be arbitrary, for example, it doesn’t need to be an InfluxDB data directory.

[ -v ]

Enables verbose logging. Confirms a file is being verified and displays progress every 5 million tombstone entries.

[ -vv ]

Enables very verbose logging. Displays progress for every series key and time range in the tombstone files. Timestamps are displayed in nanoseconds since the Epoch (1970-01-01T00:00:00Z).

[ -vvv ]

Enables very very verbose logging. Displays progress for every series key and time range in the tombstone files. Timestamps are displayed in RFC3339 format with nanosecond precision.

Note on verbose logging: Higher verbosity levels override lower levels.

Caveats

The system does not have access to the metastore when exporting TSM shards. As such, it always creates the retention policy with infinite duration and replication factor of 1. End users may want to change this prior to reimporting if they are importing to a cluster or want a different duration for retention.