Repairing the cluster
The cartridge repair command repairs a running application.
Here are several simple rules you need to know before using this command:
- Do not use
repairif you aren’t sure that it’s exactly what you need. - Always run
repairwith the--dry-runflag first. - Do not hesitate to use the
--verboseoption. - Do not use the
--forceoption if you aren’t sure that it’s exactly what you need.
Note
Please look at the
troubleshooting documentation
before using repair.
cartridge repair [subcommand]
Below is a list of the available repair commands.
list-topology
cartridge repair list-topology [flags]
Print the current topology summary. Requires no arguments.
remove-instance
cartridge repair remove-instance UUID [flags]
Remove an instance with the specified UUID from the cluster.
If the instance isn’t found, raise an error.
set-leader
cartridge repair set-leader REPLICASET-UUID INSTANCE-UUID [flags]
Set an instance as the leader of the replica set.
Raise an error in the following cases:
- There is no replica set or instance with that UUID.
- The instance doesn’t belong to the replica set.
- The instance has been disabled or expelled.
set-uri
cartridge repair set-uri INSTANCE-UUID URI-TO [flags]
Rewrite the instance’s
advertise_uri
parameter. Raise an error if the instance isn’t found or is expelled.
The following flags work with any repair subcommand:
--name(Required) Application name.
--data-dirThe directory containing the instances’ working directories.
Defaults to /var/lib/tarantool.
The following flags work with any repair command except list-topology:
--run-dirThe directory where PID and socket files are stored.
Defaults to /var/run/tarantool.
--dry-runLaunch in dry-run mode: show changes but do not apply them.
--reloadEnable instance configuration reload after the patch.
Note
The default data and run directories for repair differ from those
used by other cartridge-cli commands. This is because repair is
intended for production use, while other commands are for local development.
repair also supports global flags.
It patches cluster-wide instance configuration files that you have on your local machine.
Note that it’s not enough to apply the new configuration, the instance has to reload it.
Although repair was created for production use, it can still be applied in
local development. The command requires to specify --name, the application name.
Also, remember that the default data directory is /var/lib/tarantool and
the default run directory is /var/run/tarantool.
To specify other directories, use the data-dir and --run-dir options correspondingly
or provide the paths in the
configuration file.
In default mode, repair walks through all cluster-wide configurations
in the <data-dir>/<app-name>.* directories, patching all the configuration
files it locates.
With the --dry-run flag specified, files won’t be patched,
and you will only see the computed configuration diff.
If different instances on the local machine use different configuration files,
repair raises an error.
To patch different configuration versions independently, use the --force option.
If your application uses cartridge >= 2.0.0,
you can also run repair with the --reload flag
to reload configuration for all your instances
through the console sockets in the run directory.
Make sure that you have the correct run directory specified
when you use --reload.
cartridge repair list-topology [flags]
Print the current topology summary. Requires no arguments.
remove-instance
cartridge repair remove-instance UUID [flags]
Remove an instance with the specified UUID from the cluster.
If the instance isn’t found, raise an error.
set-leader
cartridge repair set-leader REPLICASET-UUID INSTANCE-UUID [flags]
Set an instance as the leader of the replica set.
Raise an error in the following cases:
- There is no replica set or instance with that UUID.
- The instance doesn’t belong to the replica set.
- The instance has been disabled or expelled.
set-uri
cartridge repair set-uri INSTANCE-UUID URI-TO [flags]
Rewrite the instance’s
advertise_uri
parameter. Raise an error if the instance isn’t found or is expelled.
The following flags work with any repair subcommand:
--name(Required) Application name.
--data-dirThe directory containing the instances’ working directories.
Defaults to /var/lib/tarantool.
The following flags work with any repair command except list-topology:
--run-dirThe directory where PID and socket files are stored.
Defaults to /var/run/tarantool.
--dry-runLaunch in dry-run mode: show changes but do not apply them.
--reloadEnable instance configuration reload after the patch.
Note
The default data and run directories for repair differ from those
used by other cartridge-cli commands. This is because repair is
intended for production use, while other commands are for local development.
repair also supports global flags.
It patches cluster-wide instance configuration files that you have on your local machine.
Note that it’s not enough to apply the new configuration, the instance has to reload it.
Although repair was created for production use, it can still be applied in
local development. The command requires to specify --name, the application name.
Also, remember that the default data directory is /var/lib/tarantool and
the default run directory is /var/run/tarantool.
To specify other directories, use the data-dir and --run-dir options correspondingly
or provide the paths in the
configuration file.
In default mode, repair walks through all cluster-wide configurations
in the <data-dir>/<app-name>.* directories, patching all the configuration
files it locates.
With the --dry-run flag specified, files won’t be patched,
and you will only see the computed configuration diff.
If different instances on the local machine use different configuration files,
repair raises an error.
To patch different configuration versions independently, use the --force option.
If your application uses cartridge >= 2.0.0,
you can also run repair with the --reload flag
to reload configuration for all your instances
through the console sockets in the run directory.
Make sure that you have the correct run directory specified
when you use --reload.
cartridge repair remove-instance UUID [flags]
Remove an instance with the specified UUID from the cluster. If the instance isn’t found, raise an error.
set-leader
cartridge repair set-leader REPLICASET-UUID INSTANCE-UUID [flags]
Set an instance as the leader of the replica set.
Raise an error in the following cases:
- There is no replica set or instance with that UUID.
- The instance doesn’t belong to the replica set.
- The instance has been disabled or expelled.
set-uri
cartridge repair set-uri INSTANCE-UUID URI-TO [flags]
Rewrite the instance’s
advertise_uri
parameter. Raise an error if the instance isn’t found or is expelled.
The following flags work with any repair subcommand:
--name(Required) Application name.
--data-dirThe directory containing the instances’ working directories.
Defaults to /var/lib/tarantool.
The following flags work with any repair command except list-topology:
--run-dirThe directory where PID and socket files are stored.
Defaults to /var/run/tarantool.
--dry-runLaunch in dry-run mode: show changes but do not apply them.
--reloadEnable instance configuration reload after the patch.
Note
The default data and run directories for repair differ from those
used by other cartridge-cli commands. This is because repair is
intended for production use, while other commands are for local development.
repair also supports global flags.
It patches cluster-wide instance configuration files that you have on your local machine.
Note that it’s not enough to apply the new configuration, the instance has to reload it.
Although repair was created for production use, it can still be applied in
local development. The command requires to specify --name, the application name.
Also, remember that the default data directory is /var/lib/tarantool and
the default run directory is /var/run/tarantool.
To specify other directories, use the data-dir and --run-dir options correspondingly
or provide the paths in the
configuration file.
In default mode, repair walks through all cluster-wide configurations
in the <data-dir>/<app-name>.* directories, patching all the configuration
files it locates.
With the --dry-run flag specified, files won’t be patched,
and you will only see the computed configuration diff.
If different instances on the local machine use different configuration files,
repair raises an error.
To patch different configuration versions independently, use the --force option.
If your application uses cartridge >= 2.0.0,
you can also run repair with the --reload flag
to reload configuration for all your instances
through the console sockets in the run directory.
Make sure that you have the correct run directory specified
when you use --reload.
cartridge repair set-leader REPLICASET-UUID INSTANCE-UUID [flags]
Set an instance as the leader of the replica set. Raise an error in the following cases:
- There is no replica set or instance with that UUID.
- The instance doesn’t belong to the replica set.
- The instance has been disabled or expelled.
set-uri
cartridge repair set-uri INSTANCE-UUID URI-TO [flags]
Rewrite the instance’s
advertise_uri
parameter. Raise an error if the instance isn’t found or is expelled.
The following flags work with any repair subcommand:
--name(Required) Application name.
--data-dirThe directory containing the instances’ working directories.
Defaults to /var/lib/tarantool.
The following flags work with any repair command except list-topology:
--run-dirThe directory where PID and socket files are stored.
Defaults to /var/run/tarantool.
--dry-runLaunch in dry-run mode: show changes but do not apply them.
--reloadEnable instance configuration reload after the patch.
Note
The default data and run directories for repair differ from those
used by other cartridge-cli commands. This is because repair is
intended for production use, while other commands are for local development.
repair also supports global flags.
It patches cluster-wide instance configuration files that you have on your local machine.
Note that it’s not enough to apply the new configuration, the instance has to reload it.
Although repair was created for production use, it can still be applied in
local development. The command requires to specify --name, the application name.
Also, remember that the default data directory is /var/lib/tarantool and
the default run directory is /var/run/tarantool.
To specify other directories, use the data-dir and --run-dir options correspondingly
or provide the paths in the
configuration file.
In default mode, repair walks through all cluster-wide configurations
in the <data-dir>/<app-name>.* directories, patching all the configuration
files it locates.
With the --dry-run flag specified, files won’t be patched,
and you will only see the computed configuration diff.
If different instances on the local machine use different configuration files,
repair raises an error.
To patch different configuration versions independently, use the --force option.
If your application uses cartridge >= 2.0.0,
you can also run repair with the --reload flag
to reload configuration for all your instances
through the console sockets in the run directory.
Make sure that you have the correct run directory specified
when you use --reload.
cartridge repair set-uri INSTANCE-UUID URI-TO [flags]
Rewrite the instance’s advertise_uri parameter. Raise an error if the instance isn’t found or is expelled.
The following flags work with any repair subcommand:
--name |
(Required) Application name. |
--data-dir |
The directory containing the instances’ working directories.
Defaults to /var/lib/tarantool. |
The following flags work with any repair command except list-topology:
--run-dir |
The directory where PID and socket files are stored.
Defaults to /var/run/tarantool. |
--dry-run |
Launch in dry-run mode: show changes but do not apply them. |
--reload |
Enable instance configuration reload after the patch. |
Note
The default data and run directories for repair differ from those
used by other cartridge-cli commands. This is because repair is
intended for production use, while other commands are for local development.
repair also supports global flags.
It patches cluster-wide instance configuration files that you have on your local machine. Note that it’s not enough to apply the new configuration, the instance has to reload it.
Although repair was created for production use, it can still be applied in
local development. The command requires to specify --name, the application name.
Also, remember that the default data directory is /var/lib/tarantool and
the default run directory is /var/run/tarantool.
To specify other directories, use the data-dir and --run-dir options correspondingly
or provide the paths in the
configuration file.
In default mode, repair walks through all cluster-wide configurations
in the <data-dir>/<app-name>.* directories, patching all the configuration
files it locates.
With the --dry-run flag specified, files won’t be patched,
and you will only see the computed configuration diff.
If different instances on the local machine use different configuration files,
repair raises an error.
To patch different configuration versions independently, use the --force option.
If your application uses cartridge >= 2.0.0,
you can also run repair with the --reload flag
to reload configuration for all your instances
through the console sockets in the run directory.
Make sure that you have the correct run directory specified
when you use --reload.