cockroach debug tsdump

The cockroach debug tsdump connects to your cluster and collects timeseries diagnostic data from each active node (inactive nodes are not included). This includes both current and historical runtime metrics for your cluster, including those exposed in the pages as well as internal metrics.

cockroach debug tsdump is mostly used in tandem with the command to gather diagnostic data during escalations to Cockroach Labs support. Follow the steps in this procedure to gather and prepare the timeseries diagnostic data and prepare it for transit to Cockroach Labs.

Subcommands

While the cockroach debug command has a few subcommands, users are expected to use only the , , , , , and subcommands.

We recommend using the and subcommands only when directed by the .

The other debug subcommands are useful only to Cockroach Labs. Output of debug commands may contain sensitive or secret information.

Synopsis

$ cockroach debug tsdump {flags} > {dump file destination}

The following flags must apply to an active CockroachDB node. If no nodes are live, you must .

Flags

The debug tsdump subcommand supports the following general-use, client connection, and logging flags.

General

Flag	Description
--format	The output format to write the collected diagnostic data. Valid options are text, csv, tsv, raw. Default:raw
--from	The oldest timestamp to include (inclusive), in the format YYYY-MM-DD [HH:MM[:SS]]. Default:0001-01-01 00:00:00
--metrics-list-file	Path to a text file containing metric names or regex patterns to include in the dump (one per line). The prefixes cr.node., cr.store., and cockroachdb. are automatically stripped if present. When specified, only matching metrics are included. Blank lines and comment lines (starting with #) are ignored. Useful for scenario-specific investigations (for example, contention, latency, or replication) or when you need a targeted subset of metrics. Refer to example Generate a tsdump with specific metrics. Note: Cannot be used with --non-verbose.
--non-verbose	Dump only metrics tagged as ESSENTIAL or SUPPORT. This provides a curated set of metrics needed for most escalations while significantly reducing file size and noise. Ideal for standard support escalations, routine health checks, and cases where you want to minimize output size. When this flag is not specified, all metrics are dumped (default behavior). Refer to example Generate a tsdump with only essential and support metrics. Note: Cannot be used with --metrics-list-file.
--output	Path to a file for the command to write its output to. When this flag is not specified, the command writes to stdout.
--to	The newest timestamp to include (inclusive), in the format YYYY-MM-DD [HH:MM[:SS]]. Default: Current timestamp plus 29 hours

Available resolutions

A debug tsdump includes all the available resolutions (10s and 30m) of the raw time series values in the time range set using the --from and --to flags:

• Time series values will be dumped with a 10 seconds resolution if the supplied time range is within the 10s TTL set by the cluster setting (default: 10 days).
• Time series values will be dumped with a 30 minutes resolution if the supplied time range is older than the 10s TTL back to the 30m TTL set by the cluster setting (default: 90 days).

For example, if today was 2024-11-15 and --to was set to the current timestamp (2024-11-15), and --from was set to 15 days before the current timestamp (2024-11-01), the debug tsdump would include:

• Time series values with a 10 seconds resolution for 2024-11-05 to 2024-11-15 (within the 10s TTL of 10 days).
• Time series values with a 30 minutes resolution for 2024-11-01 to 2024-11-04 (outside the 10s TTL of 10 days, but within the 30s TTL of 90 days).

Client connection

Flag	Description
	--cert-principal-map	A comma-separated list of : mappings. This allows mapping the principal in a cert to a DB principal such as node or root or any SQL user. This is intended for use in situations where the certificate management system places restrictions on the Subject.CommonName or SubjectAlternateName fields in the certificate (e.g., disallowing a CommonName like node or root). If multiple mappings are provided for the same “, the last one specified in the list takes precedence. A principal not specified in the map is passed through as-is via the identity function. A cert is allowed to authenticate a DB principal if the DB principal name is contained in the mapped CommonName or DNS-type SubjectAlternateName fields.
--certs-dir	The path to the containing the CA and client certificates and client key. Env Variable:COCKROACH\_CERTS\_DIR Default:${HOME}/.cockroach-certs/
--cluster-name	The cluster name to use to verify the cluster’s identity. If the cluster has a cluster name, you must include this flag. For more information, see .
--disable-cluster-name-verification	Disables the cluster name check for this command. This flag must be paired with --cluster-name. For more information, see .
--host	The server host and port number to connect to. This can be the address of any node in the cluster. Env Variable:COCKROACH\_HOST Default:localhost:26257
--insecure	Use an insecure connection. Env Variable:COCKROACH\_INSECURE Default:false
--url	A to use instead of the other arguments. To convert a connection URL to the syntax that works with your client driver, run . Env Variable:COCKROACH\_URL Default: no URL
--user	The SQL user that will own the client session. Valid values are root and debug\_user. Default:root

Logging

By default, this command logs messages to stdout. If you need to troubleshoot this command’s behavior, you can .

Examples

Generate a tsdump gob file

Generate the tsdump gob file for an insecure CockroachDB cluster:

$ cockroach debug tsdump --format=raw --insecure > tsdump.gob

Generate the tsdump gob file for a secure CockroachDB cluster:

$ cockroach debug tsdump --format=raw --certs-dir=${HOME}/.cockroach-certs/ > tsdump.gob

Secure examples assume you have the appropriate certificates in the default certificate directory, ${HOME}/.cockroach-certs/. See the documentation for more information.

Generate a tsdump gob file and compress using gzip

Generate a tsdump gob file for an insecure CockroachDB cluster, and compress using gzip in preparation to send to Cockroach Labs for troubleshooting. Your server must have gzip installed:

$ cockroach debug tsdump --format=raw --insecure > tsdump.gob
gzip tsdump.gob

Generate a tsdump gob file for a secure CockroachDB cluster, and compress using gzip in preparation to send to Cockroach Labs for troubleshooting. Your server must have gzip installed:

$ cockroach debug tsdump --format=raw --certs-dir=${HOME}/.cockroach-certs/ > tsdump.gob
gzip tsdump.gob

Secure examples assume you have the appropriate certificates in the default certificate directory, ${HOME}/.cockroach-certs/. See the documentation for more information.

Generate a tsdump gob file with a custom timestamp range

Generate a tsdump gob file specifying a custom timestamp range to limit the data collection to a specific interval. This is useful for reducing the size of the resulting gob file if the data needed to troubleshoot falls within a known timestamp range:

$ cockroach debug tsdump --format=raw --from='2023-01-10 01:00:00' --to='2023-01-20 23:59:59' > tsdump.gob

Generate a tsdump with only essential and support metrics

Use the --non-verbose flag to dump only metrics tagged as ESSENTIAL and SUPPORT. This option is ideal for:

• Standard support escalations
• Routine health checks
• Cases where you want to minimize file size and collection time

$ cockroach debug tsdump --format=raw --non-verbose --insecure > tsdump.gob

For a secure cluster:

$ cockroach debug tsdump --format=raw --non-verbose --certs-dir=${HOME}/.cockroach-certs/ > tsdump.gob

Generate a tsdump with specific metrics

Use the --metrics-list-file flag to include only specific metrics in a tsdump by providing a file with metric names or regular expression patterns. This option is ideal for:

• Scenario-specific investigations (for example, contention, latency, or replication issues)
• Following runbooks for specific classes of issues
• Investigating a targeted subset of metrics

First, create a text file with the metrics you want to include. You can specify exact metric names or use regex patterns. Only metrics whose names match the patterns in the file are included in the output. Blank lines and comment lines (starting with #) are ignored:

$ cat > metrics.txt <<EOF
# Example metrics list for investigating changefeed and transaction issues
changefeed.commit_latency
sql.txn.aborts

# Match all logical_replication metrics
logical_replication\..*

# Match all metrics containing "capacity"
.*capacity.*

# Match all sql metrics
sql.*
EOF

Then generate the tsdump with only the specified metrics:

$ cockroach debug tsdump --format=raw --metrics-list-file=metrics.txt --insecure > tsdump.gob

For a secure cluster:

$ cockroach debug tsdump --format=raw --metrics-list-file=metrics.txt --certs-dir=${HOME}/.cockroach-certs/ > tsdump.gob

See also