run command creates a new container from an image or starts the container with the same name and images.
Usage: run <options> [<image>][+skin(color)] [<parameters>...] <options> available: -- Parameters after -- are passed directly to the container process -a, --attach Attach to stdin, stdout, and stderr of the container. This setting is not saved in the container. --ad-domain-allow=VALUE Allow execution from the Active Directory domain --ad-domain-deny=VALUE Disallow execution from the Active Directory domain --ad-group-allow=VALUE Allow execution for members of the Active Directory group --ad-group-deny=VALUE Disallow execution for members of the Active Directory group --admin Run the container with administrative permissions --app=VALUE Run the application specified by an URI -d, --detach Run the container in the background --diagnostic Enable diagnostic logging --disable=VALUE Disable the specified Turbo VM setting. This setting is not saved in the container. -e, --env=VALUE Set an environment variable inside the container. This setting is not saved in the container. --enable=VALUE Enable the specified Turbo VM setting. This setting is not saved in the container. --enable-sync Enable container synchronization. This setting is not saved in the container. --env-file=VALUE Read in a line delimited file of environment variables. This setting is not saved in the container. --format=VALUE Use the specified format for output. Supported values: json --hide-drive=VALUE Hide specified drives. This setting is not saved in the container. Format: <*|V:|-V:>[,...] --hosts=VALUE Add an entry to the virtual /etc/hosts file (<redirect>:<name>). This setting is not saved in the container. -i, --isolate=VALUE Set isolation level: full, write-copy, merge or merge-user. --link=VALUE Add link to another container (<container>:<alias>). This setting is not saved in the container. --mount=VALUE Mount a host folder into the container. This setting is not saved in the container. Format: [other-container:]SourceFolder[=TargetFolder] --mount-if=VALUE Mount a host folder into the container but does not fail if the mount point doesn't exist. This setting is not saved in the container. Format: [other-container:]SourceFolder[=TargetFolder] -n, --name=VALUE Name of created container --network=VALUE Run container in specified named network. This setting is not saved in the container. --no-pull Uses local images to run the container if possible. If not present, will pull from the hub. --proxy-password=VALUE Password used to connect to the proxy server. It is not required when using Tnlr. This setting is not saved in the container. --proxy-server=VALUE Proxy or Tnlr server (Tnlr requires TLS). This setting is not saved in the container. Format: [socks5|http|https|tnlr]://proxy-address:port. --proxy-targets=VALUE A list of addresses (separated by semicolons) which will be proxied. If not set, all traffic will go through the proxy. The address could be either a DNS name or an IP address. --proxy-username=VALUE Username used to connect to the proxy server. It is not required when using Tnlr. This setting is not saved in the container. --pull Pulls base images from hub before running, if they exist --route-add=VALUE Add route mapping. Supported protocols: ip, pipe, tcp, and udp. This setting is not saved in the container. --route-block=VALUE Block specified route or protocol. Supported protocols: ip, tcp, and udp. This setting is not saved in the container. --route-file=VALUE Read in a INI file of routing configuration. This setting is not saved in the container. --set-startup-file=VALUE Override the default startup file permanantly --set-startup-verb=VALUE Override the default startup verb permanantly --show-window=VALUE Controls how window is shown. Supported values: max, default --startup-file=VALUE Override the default startup file. This setting is not saved in the container. --startup-verb=VALUE Override the default startup verb. This setting is not saved in the container. --stream Enable streaming of the image if available --temp Remove container when it exits --trigger=VALUE Execute named group of startup files. This setting is not saved in the container. --use-dll-cache Use image DLL cache as created by the precache command. This setting is not saved in the container. --using=VALUE Use specified images as a temporary dependency --vm=VALUE Use the specified Turbo VM version for execution -w, --working-dir=VALUE Set the initial working directory inside the container. This setting is not saved in the container. --wait-after-error Leave process open after error --wait-after-exit Leave process open after it exits
run can be used to specify multiple images by separating each image with a comma. If the same file, registry entry, or environment variable exists in multiple images, then the one from whichever image was specified last will win the conflict and be used in the virtual environment. Virtual machine settings are taken from the last specified image. Due to this "layering" approach, it is a good practice to specify images with newer versions of applications or libraries after images with older versions.
# Create a container with the apache/apache image > turbo run apache/apache # Create a container with apache and mysql > turbo run apache/apache,mysql/mysql # Create a container with .NET 3 and 4 > turbo run microsoft/dotnet:3.5.1,microsoft/dotnet:4.0.3
To use images temporarily, without committing them to the final image, use the
--using switch. This is handy for a tool like 7zip and Git that may only needed during the build process. Using images will be used with the highest overriding priority of files and registry keys.
# Create a container using git temporarily to get a project # Note that the using images are layered last, which will # means its files and registries will override all others. > turbo run --using git/git clean Using VM 20.5.1515 from local Using image clean:30 from local Using image git:2.23.0.windows.1 from local Running new container clean#0x3842xd # Clone a git project (0x3842xd) C:\> git clone https://github.com/JodaOrg/joda-time.git C:\root # Build project... # Exit and commit image (0x3842xd) C:\> exit # Git will not be part of the container after shutdown
Containers are started with the startup file specified in the first passed image that is not a modifier layer. A modifier layer is an image that has no startup file or auto-start services defined. Regardless of its position on the command line, it is never used as the entry point to the container.
# Chrome will be used as the startup file image and virtual machine settings. # However, the files in Firefox will be layered on top (overriding) any files # or registry keys in Chrome. > turbo run google/chrome,mozilla/firefox
If a startup file is not set in the base image then
cmd.exe /k is used.
# Default startup file is used to start container > turbo run oracle/jdk # Override the startup file to use the command prompt > turbo run --startup-file=cmd.exe oracle/jdk
When passing arguments to a startup file or command, we recommend separating these arguments from the rest of the command with a
--. Arguments specified after the
-- mark are passed directly to the startup file/command.
-- mark is not used, any argument matching a
run command flag will be interpreted by Turbo which may lead to unexpected behavior.
# Turbo will interpret the /d flag and execute a container in detached mode > turbo run spoonbrew/clean /d # /d flag is passed to cmd.exe, disabling execution of AutoRun commands from the registry > turbo run spoonbrew/clean -- /d
A container's standard streams (stdin/out/err) can be redirected to either the current command prompt or the background using the
# Redict standard streams to current command prompt > turbo run -a <image> # Detach the container from the native prompt > turbo run -d <image>
Detaching from a container will allow further work to be done in the native prompt while the container is running.
The initial working directory for the container can be set with the
workdir instruction or the
-w flag. The current directory will be used if
workdir was not specified and no
--startup-file parameter was provided when building the image.
# By default, a container's working directory matches the host's working directory C:\Users> turbo run git/git (0x3842xd) C:\Users> # This sets the working directory to the root of the C drive C:\Users> turbo run -w="C:\" git/git (0x3842xd) C:\>
Containerized applications can be distinguished from normal apps with skin layering. Passing
+skin(color) switch after image names pulls skin layer from hub. Skin layer paints colored border around all containerized application windows.
# Opens detached, containerized notepad with blue border around its window turbo run --startup-file=notepad -d clean+skin(blue)
Turbo VM settings can be enabled or disabled with the
--disable flags, respectively. For a list of Turbo VM settings, see VM Settings section of the documentation.
--diagnostic flag is used, the container will generate diagnostic logs that detail all of the operations that occur within the container. These diagnostic logs can later be viewed using the
turbo logs command and be used to troubleshoot errors and configuration issues.
Please note that
turbo.exe always runs outside of the container on the host even if executed from within the container.
Environment variables from the host system will be seen inside the container, unless they were isolated or set image in which the image's value will be used over the host systems. Add or override them in the container with the
--env-file flags. Setting evironment variables in the container will persist them in the container but not on the host system regardless of the isolation mode.
# Add environment variable 'foo' with value 'bar' > turbo run -e=foo=bar <image> # Specify multiple env vars with multiple flags > turbo run -e=foo=bar -e=x=2 <image>
If your container requires several environment variables then we recommend creating an env-file. An env-file is a line-delimited text file that lists all the environment variables to add to the container. The example file below lists 3 environment variables:
foo=bar utensil=turbo my-var=10
Environment variables are expanded on the host system before they are added to the container.
> echo %PATH% C:\Windows\system32;C:\Windows; > turbo run -e=%PATH%;C:\Users <image> (2fedfja3) > echo %PATH% C:\Windows\system32;C:\Windows;C:\Users
By default, containers run in the host network, meaning that any services exposed by a container can be accessible to the outside world just as if the application was running natively on the host. However, it is possible to run containers in virtualized network environments by specifying a network name other than "host" with the
--network flag. Running a container in virtualized network environment prevents it from exposing services to the outside world (unless
--route-add flag is used appropriately) while allowing for easy communication between containers running in the same virtualized network environment. In a virtual network, containers can connect to each other using their names as specified with the
--name flag if there was any or auto-generated from the image name otherwise.
# Launch a new container in the host network context (the default) > turbo run --network=host <image> # Launch two containers in a "mynet" virtual network > turbo run -d --network=mynet --name=web <image> web#88e3bb0e > turbo run -d --network=mynet myself/webbrowser http://web webbrowser#dd73e48a # The former will accessible by its name "web" within the network, # and the latter by its auto-generated name: "webbrowser"
Note: When connecting, always use the container name and not the network name. After all, what should your application connect to if there were two separate containers exposing the same services on the same virtual network if you connected by network name instead of container name?
All network operations (opening/closing ports, for example) are passed through to the local machine when running in the host network context. To remap container ports to other ports on the local machine, use the
--route-add flag. This flag also works when running in a virtualized network environment (by specifying the
# Map container tcp port 8080 to local port 80 > turbo run --route-add=tcp://8080:80 <image> # Map udp traffic on container port 8080 to local port 80 > turbo run --route-add=udp://8080:80 <image> # Map container tcp port 80 to random port on local machine # The random port can be later queried using the netstat command > turbo run --route-add=tcp://80:0 <image>
The default policy of allowing containers to bind to any port on the local machine can be changed with the
--route-block flag. It isolates all services bound to container ports on specified protocols (tcp or udp). They can only be opened using the
# Isolate all tcp services of a container > turbo run --route-block=tcp <image> # Isolate all tcp and udp services, but allow container tcp port 3486 # be bound to port 80 on local machine > turbo run --route-block=tcp,udp --route-add=tcp://3486:80 <image>
If you decided to not expose any services running in a container to the public by specifying the
--route-block flag and not
--route-add, you may still want to be able to connect to the services in your container from another container on the same machine. Although this is best achieved by running the containers in the same virtual network using the
--network flag, container linking can be used for this purpose as well.
When creating a container with the
turbo new command, you can use the
--link flag to link it to any existing containers and the new container will be able to connect to any services exposed by the linked containers. Such connection creates a parent-child relationship where the newly created container is the parent.
With each link, an alias name must be specified. Name resolution overrides are added to the parent container so it can refer to its children by these names. Note how with container links the name that a container will use to refer to another container is defined by the former (the parent) using a parameter, instead of by the name of the container as is the case with virtual networks (the
Container links also work between containers running in different virtual networks.
First create two containers, each exposing web sites on private port 80, but with no services exposed outside the containers. Run them in detached mode.
> turbo run --route-block=tcp,udp -d <image> image#05bf1aa4 > turbo run --route-block=tcp,udp -d <image> image#94a38820
Then create a web browser container linked to the previously created containers.
> turbo run --link=05bf:web1 --link=94a3:web2 myself/webbrowser http://web1 http://web2
You will be able to browse websites served by the linked containers even though they are not publically available.
Controlling Outbound Traffic
--route-block not only provide a way to create rules that apply to inbound network traffic with the
udp protocols, but also rules that apply to outbound network traffic. For the outbound rules, the
ip protocol is used. The rules can be implemented using a whitelist or a blacklist approach. It is also possible to reroute traffic from one IP address/host to another, effectively defining an IP address alias.
Routes can be defined using IPv4, IPv6 addresses, or based on hostnames. Note however that you cannot specify a host name on the right side of a
--route-add mapping since the result would be ambiguous if the host name resolved to multiple IP addresses.
If your container requires several routing rules then we recommend creating a route-file. A route-file is an INI based, line-delimited text file that lists all the routing rules to add to the container. It can be added with
Create a PuTTY container with all outbound access blocked except to IP address 10.0.0.34 (whitelist approach):
> turbo run --route-block=ip --route-add=ip://10.0.0.34 putty
In addition to the above, reroute all traffic to 18.104.22.168 to 10.0.0.34, making it possible to connect to host at 10.0.0.34 typing address 22.214.171.124 in PuTTY:
> turbo run --route-block=ip --route-add=ip://10.0.0.34 --route-add=ip://126.96.36.199:10.0.0.34 putty
It is also possible to use IP ranges using the CIDR notation. The following command allows PuTTY in the container to connect only to hosts in the 192.168.1.0/24 network:
> turbo run --route-block=ip --route-add=ip://192.168.1.0/24 putty
To disallow the app to connect to a set of specific IP addresses (blacklist approach), simply specify them in the
> turbo run --route-block=ip://192.168.1.55 --route-block=ip://192.168.1.57 putty
When working with IPv6 addresses, it is necessary to enclose them in square brackets:
Block an IPv6 address:
> turbo run --route-block=ip://[2001:4860:4860::8888] putty
Block all IP traffic, except link local IPv6 space:
> turbo run --route-block=ip --route-add=ip://[fe80::c218:85ff:febd:5c01/64] putty
Reroute traffic to an IPv6 address to localhost:
> turbo run --route-block=ip --route-add=ip://[2001:cdba::3257:9652]:[::1] putty
To simplify working with mutliple IP addresses it is possible to use hostnames on the left side of all commands. When a hostname is specified with
--route-block, it is resolved to an IP address when the container starts, and the behavior is effectively the same as if the IP address was specified in place of the hostname. Additionally, all DNS resolves are intercepted and whenever a known hostname resolves to a previously unknown IP address, the IP address is added to the appropriate route table. This feature is what allows wildcard hostnames to work, since otherwise it would not be possible to infer the IP addresses of all possible subdomains.
For example, to run a Chrome container allowing only access to the turbo.net and blog.turbo.net domains, you can use the command:
> turbo run --route-block=ip --route-add=ip://turbo.net --route-add=ip://blog.turbo.net chrome https://turbo.net
Wildcards are supported in host name routing. So, for example, to unblock turbo.net and all of its subdomains, use the expression:
> turbo run --route-block=ip --route-add=ip://*.turbo.net chrome https://blog.turbo.net
Or, to run a Chrome container disallowing access to the facebook.com domain and all of its subdomains:
> turbo run --route-block=ip://*.facebook.com chrome
Another option is to use an INI based route-file which defines rules for blocking and allowing network traffic. The example below blocks all network traffic and then unblocks 192.168.198.0/24 and all turbo.net and spoon.net subdomains:
[ip-block] * [ip-add] 192.168.198.0/24 *.turbo.net *.spoon.net
To create a firefox container with above route-file use this command:
turbo run --route-file=c:\turbo-rules.txt firefox https://turbo.net
If a large list of hostnames is used, such as in the
turbobrowsers/block-ad-routes image, the default behavior as described above of resolving all of them to IP addresses at the start of the container would cause container startup to take too long. It can be overriden with the
PreResolveHostNames=false setting in a route file, as shown below:
[settings] PreResolveHostNames=false` [ip-block] adserver1.com adserver2.com ...
Adding Custom Name Resolution Entries
All containers use name resolution provided by the host operating system. You can add specific name resolution overrides using the
--hosts flag. The syntax is similar to that of the
hosts file of the operating system.
# Make name my-test-service resolve to whatever the name # test-service-43 resolves > turbo run --hosts=my-test-service:test-service-43 <image> # Make name mysite.net resolve to IPv4 address 127.0.0.1 and # name ipv6.mysite.net resolve to IPv6 address ::1 > turbo run --hosts=127.0.0.1:mysite.net --hosts=::1:ipv6.mysite.net <image>
Using Startup Triggers
Images can be created with TurboScript that have multiple startup files. Collections of startup files can be linked together by a trigger name and executed together.
# in turbo.me file to create "test-trigger" image... startup file ["c:\windows\system32\notepad.exe", "c:\windows\regedit.exe"] startup file doc=[("c:\windows\system32\notepad.exe", "c:\doc\welcome.txt"), ("c:\windows\system32\notepad.exe", "c:\doc\howto.txt")] # from command-prompt... # launch both notepad and regedit are launched > turbo run test-trigger # launch welcome.txt and howto.txt in notepad > turbo run test-trigger --trigger=doc
mount option provides a way to mount folders from the host into the container, giving access to resources from the host system.
The mounted folder's content is not committed to the image nor synchronized to the Turbo Hub and therefore is not available when using the
If the source folder doesn't exist, the
mount option is ignored. If the target folder doesn't exist, it is created.
Example for mounting a folder.
turbo run --mount "C:\FolderOnHostSystem=C:\FolderInContainer" clean
Mounts are useful to share a cache folder, like a local Maven repository:
turbo run --mount "%USERPROFILE%\.m2=%USERPROFILE%\.m2" jdk,maven
Mounting multiple folder is done by repeating the mount parameter:
turbo run --mount "C:\Mount1=C:\InContainer1" --mount "C:\Mount2=C:\InContainer2" clean
It is also possible to mount a folder from another container:
turbo run --mount <containerid>:"C:\FolderInSourceContainer=C:\FolderInTargetContainer" clean
isolate parameter enables different levels of visibility and access from the vm to the host environment. The
full isolation setting prevents read and write to the host system and registry. This is the preferred setting if you want the vm to run like a clean, completely isolated system.
writecopy isolation setting allows the vm to read the host file system and registry but not write to the host. This is the preferred setting if you want the vm to be able to access host applications and settings, but not alter the host in any way.
merge isolation setting allows read and write access to the host system.
Note that the vm isolation setting does not override more restrictive isolation settings that already exist in the image. For example, if you created an image in Turbo Studio and set specific folders and keys to
full isolation, those settings would be preserved even if the vm isolation is set to
For applications like Gimp or Notepad++ where you want to allow the vm to edit and save files you work with to the host file system, but otherwise do not want to let the application litter the host system for example with settings stored in the file system or the registry, there is the
+merge-user isolation modifier. Used as
write-copy+merge-user, it uses merge isolation for user folders like Desktop or Documents, but keeps the base
write-copy isolation for the rest of the system, making sure that the host system is kept clean. The preferred mode is
The well-known root folders affected by the
+merge-user modifier are:
As a separate convenience feature, if the startup verb is not empty, the startup file of the container is set to merge isolation, regardless of the isolation level that it would otherwise have. This way, when executing a shell operation like opening a file on the host system through a Turbo application that has host system file associations set, it is possible for the virtualized application to access and make changes to the file. The
MergeStartupDir vm flag takes this feature one step further and sets the isolation level to merge for the whole parent folder of the startup file and all its subfolders except well-known root folders. For example, if the startup file was
C:\myproject.proj and the flag was enabled, the folder
C:\myproject-files would have merge isolation, but e.g.,
C:\Program Files, being well-known root folders, would have isolation level unchanged.
When the container stops, the exit code of startup file is displayed in decimal form.
Selecting VM version
A specific VM version can be selected by using the
--vm=version flag. If the selected version is lower than the minimum version that is required by turbo.exe, then the minimum version will be used instead.
--format=json option was passed this command will provide output in JSON format. It will contain either a
container array with information about created container or an
error object if command failed.
run command checks once a day for new image releases. Specify the
--pull option to force checking for the latest release.
Updates are download within the specified release:
turbo run firefox updates to the latest Firefox.
turbo run firefox:42 updates to the Firefox within release 42, like 42.0, 42.1 42.2 etc.
Remote vs. Local Image Usage
The behavior is to prioritize locally cached images over remote, unless the update check triggers an update of the repositories image.
# Start with an empty local image cache > turbo run firefox Downloading image firefox:3.5 from https://turbo.net/users/mozilla Pull complete # If the update check has not been done for firefox repository, run will update to the latest version > turbo run firefox Using VM 18.7.1306 from local Using image clean:26 from local Upgrading image firefox to version 61.0.2 Upgrading image firefox-base to version 61.0.2 Running new container firefox#165115f7 # Remove the created container and the latest image > turbo rm firefox Removing container firefox#165115f7 > turbo rmi firefox:61 Image mozilla/firefox:61.0.2 was removed # Running firefox again will use the local image with the older version because the update check has already been done in the previous run > turbo run firefox Using VM 18.7.1306 from local Using image clean:26 from local Using image firefox:3.5 from local Running new container firefox#9afe83e2
If the user does not want to trigger the update to the latest firefox, specify the release explicitly:
> turbo run firefox:3.5 Using VM 18.7.1306 from local Using image clean:26 from local Using image firefox:3.5 from local Running new container firefox:3.5#a524349c
Run In Existing Session
If the user issues a run or start command targeting an already running virtual application session, then the process will start under the existing virtual session.
# Open a command prompt, detached > turbo run clean -d Running new container clean#a7b31817 # Open another command prompt, using the existing session > turbo run clean Running existing container clean#a7b31817 The application was started inside a running container # Exit the second command prompt window, the turbo command unblocks
Running a process in an existing session will wait until the process exits before completing the turbo command. The wait for exit does not include child processes. Often an application's startup file such as firefox or notepad++ will immediately exit when another instance of the application is running. The turbo command will immediately exit. Persisted session data will be synchronized in the original turbo session.