FAQs and Troubleshooting
General Questions#
Administration Console#
Q: How do I access the Administration Console?
The Administration Console allows Tower instance administrators to interact with all users and organizations registered with the platform. Administrators must be identified in your Tower instance configuration files prior to instantiation of the application.
- Create a
TOWER_ROOT_USERS
environment variable (e.g. via tower.env or Kubernetes ConfigMap). - Populate the variable with a sequence of comma-delimited email addresses (no spaces).
Example:TOWER_ROOT_USERS=foo@foo.com,bar@bar.com
- If using a Tower version earlier than 21.12:
- Add the following configuration to tower.yml:
1 2 3
tower: admin: root-users: '${TOWER_ROOT_USERS:[]}'
- Add the following configuration to tower.yml:
- Restart the
cron
andbackend
containers/Deployments. - The console will now be availabe via your Profile drop-down menu.
API#
Q:I am trying to query more results than the maximum return size allows. Can I do pagination?
Yes. We recommend using pagination to fetch the results in smaller chunks through multiple API calls with the help of max
and subsequent offset
parameters. You will receive an error like below if you run into the maximum result limit.
{object} length parameter cannot be greater than 100 (current value={value_sent})
We have laid out an example below using the workflow endpoint.
1 2 3 4 5 6 7 |
|
Q: Why am I receiving a 403 HTTP Response when trying to launch a pipeline via the /workflow/launch
API endpoint?
Launch users have more restricted permissions within a Workspace than Power users. While both can launch pipelines via API calls, Launch users must specify additional values that are optional for a Power user.
One such value is launch.id
; attempting to launch a pipeline without specifying a launch.id
in the API payload is equivalent to using the "Start Quick Launch" button within a workspace (a feature only available to Power users).
If you have encountered the 403 error as a result of being a Launch user who did not provide a launch.id
, please try resolving the problem as follows:
-
Provide the launch ID to the payload sent to the tower using the same endpoint. To do this;
- Query the list of pipelines via the
/pipelines
endpoint. Find thepipelineId
of the pipeline you intend to launch. - Once you have the
pipelineId
, call the/pipelines/{pipelineId}/launch
API to retrieve the pipeline'slaunch.id
. - Include the
launch.id
in your call to the/workflow/launch
API endpoint (see example below).1 2 3 4 5 6 7 8 9 10
{ "launch": { "id": "Q2kVavFZNVCBkC78foTvf", "computeEnvId": "4nqF77d6N1JoJrVrrgB8pH", "runName": "sample-run", "pipeline": "https://github.com/sample-repo/project", "workDir": "s3://myBucketName", "revision": "main" } }
- Query the list of pipelines via the
-
If a launch id remains unavailable to you, upgrade your user role to 'Maintain' or higher. This will allow you to execute quick launch-type pipeline invocations.
Common Errors#
Q: After following the log-in link, why is my screen frozen at /auth?success=true
?
Starting with v22.1, Tower Enterprise implements stricter cookie security by default and will only send an auth cookie if the client is connected via HTTPS. The lack of an auth token will cause HTTP-only log-in attempts to fail (thereby causing the frozen screen).
To remediate this problem, set the following environment variable TOWER_ENABLE_UNSAFE_MODE=true
.
Q: "Unknown pipeline repository or missing credentials" error when pulling from a public Github repository?
Github imposes rate limits on repository pulls (including public repositories), where unauthenticated requests are capped at 60 requests/hour and authenticated requests are capped at 5000/hour. Tower users tend to encounter this error due to the 60 request/hour cap.
To resolve the problem, please try the following:
- Ensure there is at least one Github credential in your Workspace's Credentials tab.
- Ensure that the Access token field of all Github Credential objects is populated with a Personal Access Token value, NOT a user password. (Github PATs are typically several dozen characters long and begin with a
ghp_
prefix; example:ghp_IqIMNOZH6zOwIEB4T9A2g4EHMy8Ji42q4HA
) - Confirm that your PAT is providing the elevated threshold and transactions are being charged against it:
1
`curl -H "Authorization: token ghp_LONG_ALPHANUMERIC_PAT" -H "Accept: application/vnd.github.v3+json" https://api.github.com/rate_limit`
Q: "Unexpected error sending mail ... TLS 1.0 and 1.1 are not supported. Please upgrade/update your client to support TLS 1.2" error?
Some mail services, including Microsoft, have phased out support for TLS 1.0 and 1.1. Tower Enterprise, however, is based on Java 11 (Amazon Coretto) and does not use TLSv1.2 by default. As a result, an encryption error will occur when Tower tries to send email even if you have configured your mail.smtp.starttls
settings to be true
.
To fix the problem, use this JDK environment variable to force the usage of TLSv1.2 by default:
1 |
|
Q: "Row was updated or deleted by another transaction (or unsaved-value mapping was incorrect)" error.
This error can occur if incorrect configuration values are assigned to the backend
and cron
containers' MICRONAUT_ENVIRONMENTS
environment variable. You may see other unexpected system behaviour like two exact copies of the same Nextflow job be submitted to the Executor for scheduling.
Please verify the following:
- The
MICRONAUT_ENVIRONMENTS
environment variable associated with thebackend
container:- Contains
prod,redis,ha
- Does not contain
cron
- Contains
- The
MICRONAUT_ENVIRONMENTS
environment variable associated with thecron
container:- Contains
prod,redis,cron
- Does not contain
ha
- Contains
- You do not have another copy of the
MICRONAUT_ENVIRONMENTS
environment variable defined elsewhere in your application (e.g. a tower.env file or Kubernetes ConfigMap). - If you are using a separate container/pod to execute migrate-db.sh, there is no
MICRONAUT_ENVIRONMENTS
environment variable assigned to it.
Q: "No such variable" error.
This error can occur if you execute a DSL 1-based Nextflow workflow using Nextflow 22.03.0-edge or later.
Compute Environments#
Q: Can the name of a Compute Environment created in Tower contain special characters?
No. Tower version 21.12 and later do not support the inclusion of special characters in the name of Compute Environment objects.
Q: How do I set NXF_OPTS values for a Compute Environment?
This depends on your Tower version:
- For v22.1.1+, specify the values via the Environment variables section of the "Add Compute Environment" screen.
-
For versions earlier than v22.1.1, specify the values via the Staging options > Pre-run script textbox on the "Add Compute Environment" screen. Example:
export NXF_OPTS="-Xms64m -Xmx512m"
Containers#
Q: Can I use rootless containers in my Nextflow pipelines?
Most containers use the root user by default. However, some users prefer to define a non-root user in the container in order to minimize the risk of privilege escalation. Because Nextflow and its tasks use a shared work directory to manage input and output data, using rootless containers can lead to file permissions errors in some environments:
1 |
|
As of Tower 22.1.0 or later, this issue should not occur when using AWS Batch. In other situations, you can avoid this issue by forcing all task containers to run as root. To do so, add one of the following snippets to your Nextflow configuration:
1 2 3 4 5 6 7 8 |
|
Databases#
Q: Help! I upgraded to Tower Enterprise 22.2.0 and now my database connect is failing.
Tower Enterprise 22.2.0 introduced a breaking change whereby the TOWER_DB_DRIVER
is now required to be org.mariadb.jdbc.Driver
.
Clients who use Amazon Aurora as their database solution may encounter a java.sql.SQLNonTransientConnectionException: ... could not load system variables
error, likely due to a known error tracked within the MariaDB project.
Please modify Tower Enterprise configuration as follows to try resolving the problem:
- Ensure your
TOWER_DB_DRIVER
uses the specified MariaDB URI. - Modify your
TOWER_DB_URL
to:TOWER_DB_URL=jdbc:mysql://YOUR_DOMAIN:YOUR_PORT/YOUR_TOWER_DB?usePipelineAuth=false&useBatchMultiSend=false
Datasets#
Q: Why are uploads of Datasets via direct calls to the Tower API failing?
When uploading Datasets via the Tower GUI or CLI, some steps are automatically done on your behalf. Clients wishing to upload Datasets via direct calls to the API are required to undertake a few additional steps: * [ ] 1. Explicitly define the MIME type of the file they are uploading. 2. Make two calls to the API: 1. Create a Dataset object 2. Upload the samplesheet to the Dataset object.
Example:
1 2 3 4 5 |
|
Tip
You can also use the tower-cli to upload the dataset to a particular workspace.
1 |
|
Q: Why is my uploaded Dataset not showing in the Tower Launch screen input field dropdown?
When launching a Nextflow workflow from the Tower GUI, the input
field dropdown will only show Datasets whose mimetypes match the rules specified in the associated nextflow_schema.json
file. If your Dataset has a mimetype different than what the pipeline expects, Tower will not present the file.
Example: The default nf-core RNASeq pipeline specifies that only files with a csv
mimetype should be provided as an input file. If you created a Dataset of mimetype tsv
, it would not appear as an input filed dropdown option.
Q: Can an input file mimetype restriction be added to the nextflow_schema.json file generated by the nf-core pipeline schema builder tool?
As of August 2022, it is possible to add a mimetype restriction to the nextflow_schema.json file generated by the nf-core schema builder tool but this must occur manually after generation, not during. Please refer to this RNASeq example to see how the mimetype
key-value pair should be specified.
Q: Why are my datasets converted to 'application/vnd.ms-excel' data type when uploading on a browser using Windows OS?
This is a known issue when using Firefox browser with the Tower version prior to 22.2.0. You can either (a) upgrade the Tower version to 22.2.0 or higher or (b) use Chrome.
For context, the Tower will prompt the message below if you encountered this issue.
1 |
|
Healthcheck#
Q: Does Tower offer a healthcheck API endpoint?
Yes. Customers wishing to implement automated healtcheck functionality should use Tower's service-info
endpoint.
Example:
1 2 3 |
|
Logging#
Q: Can Tower enable detailed logging related to sign-in activity?
Yes. For more detailed logging related to login events, set the following environment variable: TOWER_SECURITY_LOGLEVEL=DEBUG
.
Q: Can Tower enable detailed logging related to application activites?
Yes. For more detailed logging related to application activities, set the following environment variable: TOWER_LOG_LEVEL=TRACE
.
Login#
Q: Can I completely disable Tower's email login feature?
The email login feature cannot be completely removed from the Tower login screen.
Q: How can I restrict Tower access to only a subset of email addresses?
You can restrict which emails are allowed to have automatic access to your Tower implementation via a configuration in tower.yml.
Users without automatic access will receive an acknowledgment of their login request but be unable to access the platform until approved by a Tower administration via the Administrator Console.
1 2 3 4 5 |
|
Q: Why is my OIDC redirect_url set to http instead of https?
This can occur for several reasons. Please verify the following:
- Your
TOWER_SERVER_URL
environment variable uses thehttps://
prefix. - Your
tower.yml
hasmicronaut.ssl.enabled
set totrue
. - Any Load Balancer instance that sends traffic to the Tower application is configured to use HTTPS as its backend protocol rather than TCP.
Q: Why isn't my OIDC callback working?
Callbacks could fail for many reasons. To more effectively investigate the problem:
- Set the Tower environment variable to
TOWER_SECURITY_LOGLEVEL=DEBUG
. - Ensure your
TOWER_OIDC_CLIENT
,TOWER_OIDC_SECRET
, andTOWER_OIDC_ISSUER
environment variables all match the values specified in your OIDC provider's corresponding application. - Ensure your network infrastructure allow necessary egress and ingress traffic.
Q: Why did Google SMTP start returning Username and Password not accepted
errors?
To reestablish email connectivity, please follow the instructions at https://support.google.com/accounts/answer/3466521 to provision an app password. Update your TOWER_SMTP_PASSWORD
environment variable with the app password, and restart the application.
Miscellaneous#
Q: Is my data safe?
Yes, your data stays strictly within your infrastructure itself. When you launch a workflow through Tower, you need to connect your infrastructure (HPC/VMs/K8s) by creating the appropriate credentials and compute environment in a workspace.
Tower then uses this configuration to trigger a Nextflow workflow within your infrastructure similar to what is done via the Nextflow CLI, therefore Tower does not manipulate any data itself and no data is transferred to the infrastructure where Tower is running.
Monitoring#
Q: Can Tower integrate with 3rd party Java-based Application Performance Monitoring (APM) solutions?
Yes. You can mount the APM solution's JAR file in the backend
container and set the agent JVM option via the JAVA_OPTS
env variable.
Q: Is it possible to retrieve the trace file for a Tower-based workflow run?
-
Set the following block in your
nextflow.config
:1 2 3
trace { enabled = true }
-
Add a copy command to your pipeline's Advanced options > Post-run script field:
1 2 3
# Example: Export the generated trace file to an S3 bucket # Ensure that your Nextflow head job has the necessary permissions to interact with the target storage medium! aws s3 cp ./trace.txt s3://MY_BUCKET/trace/trace.txt
Q: When monitoring pipeline execution via the Runs tab, why do I occasionally see Tower reporting "Live events sync offline"?
Nextflow Tower uses server-sent events to push real-time updates to your browser. The client must establish a connection to the Nextflow Tower server's /api/live
endpoint to initiate the stream of data, and this connection can occasionally fail due to factors like network latency.
To resolve the issue, please try reloading the UI to reinitiate the client's connection to the server. If reloading fails to resolve the problem, please contact Seqera Support for assistance with webserver timeout settings adjustments.
Nextflow Configuration#
Q: Can a repository's nextflow_schema.json
support multiple input file mimetypes?
No. As of April 2022, it is not possible to configure an input field (example) to support different mime types (e.g. a text/csv
-type file during one execution, and a text/tab-separated-values
file in a subsequent run).
Q: Why are my --outdir
artefacts not available when executing runs in a cloud environment?
As of April 2022, Nextflow resolves relative paths against the current working directory. In a classic grid HPC, this normally corresponds to a subdirectory of the user's $HOME directory. In a cloud execution environment, however, the path will be resolved relative to the container file system meaning files will be lost when the container is termination. See here for more details.
Tower Users can avoid this problem by specifying the following configuration in the Advanced options > Nextflow config file configuration textbox: params.outdir = workDir + '/results
. This will ensure the output files are written to your stateful storage rather than ephemeral container storage.
Q: Can Nextflow be configured to ignore a Singularity cache?
Yes. To ignore the Singularity cache, add the following configuration item to your workflow: process.container = 'file:///some/singularity/image.sif'
.
Q: Why does Nextflow fail with a WARN: Cannot read project manifest ... path=nextflow.config
error message?
This error can occur when executing a pipeline where the source git repository's default branch is not populated with main.nf
and nextflow.config
files, regardles of whether the invoked pipeline is using a non-default revision/branch (e.g. dev
).
Current as of May 16, 2022, there is no solution for this problem other than to create blank main.nf
and nextflow.config
files in the default branch. This will allow the pipeline to run, using the content of the main.nf
and nextflow.config
in your target revision.
Q: Is it possible to maintain different Nextflow configuration files for different environments?
Yes. The main nextflow.config
file will always be imported by default. Instead of managing multiple nextflow.config
files (each customized for an environment), you can create unique environment config files and import them as their own profile in the main nextflow.config
.
Example:
1 2 3 4 5 6 7 8 9 10 11 |
|
Q: Is there a limitation to the size of the BAM files that can be uploaded to the S3 bucket?
You will see this on your log file if you encountered an error related to this:
WARN: Failed to publish file: s3://[bucket-name]
AWS have a limitation on the size of the object that can be uploaded to S3 when using the multipart upload feature. You may refer to this documentation for more information. For this specific instance, it is hitting the maximum number of parts per upload.
The following configuration are suggested to work with the above stated AWS limitation:
- Head Job CPUs = 16
- Head Job Memory = 60000
- Pre-run script = export NXF_OPTS="-Xms20G -Xmx40G"
- Update the
nextflow.config
to increase the chunk size and slow down the number of transfers.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
aws { batch { maxParallelTransfers = 5 maxTransferAttempts = 3 delayBetweenAttempts = 30 } client { uploadChunkSize = '200MB' maxConnections = 10 maxErrorRetry = 10 uploadMaxThreads = 10 uploadMaxAttempts = 10 uploadRetrySleep = '10 sec' } }
Q: Why is Nextflow forbidden to retrieve a params file from Nextflow Tower?
Ephemeral endpoints can only be consumed once. Nextflow versions older than 22.04
may try to call the same endpoint more than once, resulting in an error similar to the following:
Cannot parse params file: /ephemeral/example.json - Cause: Server returned HTTP response code: 403 for URL: https://api.tower.nf/ephemeral/example.json
.
To resolve this problem, please upgrade your Nextflow version to version 22.04.x
or later.
Q: How can I prevent Nextflow from uploading intermediate files from local scratch to my S3 work directory?
Nextflow will only unstage files/folders that have been explicitly defined as process outputs. If your workflow has processes that generate folder-type outputs, please ensure that the process also purges any intermediate files that reside within. Failure to do so will result in the intermediate files being copied as part of the task unstaging process, resulting in additional storage costs and lengthened pipeline execution times.
Q: Why do some values specified in my git repository's nextflow.config change when the pipeline is launched via Tower?
Example: The following code block is specified in your nextflow.config:
1 2 3 4 5 6 7 |
|
When the job instantiates on the AWS Batch Compute Environment, you will see that the uploadChunkSize
changed:
1 2 3 4 5 6 7 |
|
This change occurred because Tower superimposes its 10 MB default value rather than using the value specified in the nextflow.config file.
To force the Tower-invoked job to use your desired value, please add the configuration setting in the Tower Workspace Launch screen's Advanced options > Nextflow config file textbox. In the case of our example above, you would simply need to add aws.client.uploadChunkSize = 209715200 // 200 MB
.
Nextflow configuration values that are affected by this behaviour include:
- aws.client.uploadChunkSize
- aws.client.storageEncryption
Nextflow Launcher#
Q: There are several nf-launcher images available in the Seqera image registry. How can I tell which one is most appropriate for my implementation?
Your Tower implementation knows the nf-launcher image version it needs and will specify this value automatically when launching a pipeline.
If you are restricted from using public container registries, please see Tower Enterprise Release Note instructions (example) for the specific image you should use and how to set this as the default when invoking pipelines.
Q: The nf-launcher is pinned to a specific Nextflow version. How can I make it use a different release?
Each Nextflow Tower release uses a specific nf-launcher image by default. This image is loaded with a specific Nextflow version, meaning that any workflow run in the container uses this Nextflow version by default. You can force your jobs to use a newer/older version of Nextflow with any of the following strategies:
- Use the Pre-run script advanced launch option to set the desired Nextflow version. Example:
export NXF_VER=22.08.0-edge
- For jobs executing in an AWS Batch compute environment, create a custom job definition which references a different nf-laucher image.
OIDC#
Q: Can I have users seamlessly log in to Tower if they already have an active session with their OpenId Connect (OIDC) Identity Provider (IDP)?
Yes. If you are using OIDC as your authentication method, it is possible to implement a seamless login flow for your users.
Rather than directing your users to http(s)://YOUR_TOWER_HOSTNAME
or http(s)://YOUR_TOWER_HOSTNAME/login
, point the user-initiated login URL here instead: http(s)://YOUR_TOWER_HOSTNAME/oauth/login/oidc
.
If your user already has an active session already established with the IDP, they will be automatically logged into Tower rather than having to manually choose their authentication method.
Plugins#
Q: Is it possible to use the Nextflow SQL DB plugin to query AWS Athena?
Yes. As of Nextflow 22.05.0-edge, your Nextflow pipelines can query data from AWS Athena.
You must add the following configuration items to your nextflow.config
(Note: the use of secrets is optional):
1 2 3 4 5 6 7 8 9 10 11 12 13 |
|
You can then call the functionality from within your workflow.
1 2 3 |
|
For more information on the implementation, please see https://github.com/nextflow-io/nf-sqldb/discussions/5.
Repositories#
Q: Can Tower integrate with private docker registries like JFrog Artifactory?
Yes. Tower-invoked jobs can pull container images from private docker registries. The method to do so differs depending on platform, however:
- If using AWS Batch, modify your EC2 Launch Template as per these directions from AWS.
Note:- This solution requires that your Docker Engine be at least 17.07 to use
--password-stdin
. - You may need to add the following additional commands to your Launch Template depending on your security posture:
cp /root/.docker/config.json /home/ec2-user/.docker/config.json && chmod 777 /home/ec2-user/.docker/config.json
- This solution requires that your Docker Engine be at least 17.07 to use
- If using Azure Batch, please create a Container Registry-type credential in your Tower Workspace and associate it with the Azure Batch object also defined in the Workspace.
- If using Kubernetes, please use an
imagePullSecret
as per https://github.com/nextflow-io/nextflow/issues/2827.
Q: Why does my Nextflow log have a Remote resource not found
error when trying to contact the workflow repository?
This error can occur if the Nextflow head job fails to retrieve the necessary repository credentials from Nextflow Tower.
To determine if this is the case, please do the following:
- Check your Nextflow log for an entry like
DEBUG nextflow.scm.RepositoryProvider - Request [credentials -:-]
. - If the above is true, please check the protocol of the string that was assigned to your Tower instance's
TOWER_SERVER_URL
configuration value. It is possible this has been erroneously set tohttp
rather thanhttps
.
Secrets#
Q: When using secrets in Tower workflow run, the process executed with an error Missing AWS execution role arn
The ECS Agent must be empowered to retrieve Secrets from the AWS Secrets Manager. Secrets-using pipelines that are launched from Nextflow Tower and execute in an AWS Batch Compute Environment will encounter this error if an IAM Execution Role is not provided. Please see the Pipeline Secrets for remediation steps.
Q: Why do work tasks which use Secrets fail when running in AWS Batch?
Users may encounter a few different errors when executing pipelines that use Secrets, via AWS Batch:
-
If you use
nf-sqldb
version 0.4.1 or earlier and have Secrets in yournextflow.config
, you may see following error in your Nextflow Log:nextflow.secret.MissingSecretException: Unknown config secret {SECRET_NAME}
.
You can resolve this error by explicitly defining thexpack-amzn
plugin in your configuration.
Example:1 2 3 4
plugins { id 'xpack-amzn' id 'nf-sqldb' }
-
If you have two or more processes that use the same container image, but only a subset of these processes use Secrets, your Secret-using processes may fail during the initial run but succeed when resumed. This is due to an bug in how Nextflow (22.07.1-edge and earlier) registers jobs with AWS Batch.
To resolve the issue, please upgrade your Nextflow version to 22.08.0-edge. If you cannot upgrade, you can use the following as workarounds:
- Use a different container image for each process.
- Define the same set of Secrets in each process that uses the same container image.
Tower Configuration#
Q: Can a custom path be specified for the tower.yml
configuration file?
Yes. Provide a POSIX-compliant path to the TOWER_CONFIG_FILE
environment variable.
Q: Why do parts of tower.yml
not seem to work when I run my Tower implementation?
There are two reasons why configurations specified in tower.yml
are not being expressed by your Tower instance:
- There is a typo in one of the key value pairs.
- There is a duplicate key present in your file.
1 2 3 4 5 6 7 8 9 10 11
# EXAMPLE # This block will not end up being enforced because there is another `tower` key below. tower: trustedEmails: - user@example.com # This block will end up being enforced because it is defined last. tower: auth: oidc: - "*@foo.com"
Q: Do you have guidance on how to create custom Nextflow containers?
Yes. Please see https://github.com/seqeralabs/gatk4-germline-snps-indels/tree/master/containers.
Q: What DSL version does Nextflow Tower set as default for Nextflow head jobs?
As of Nextflow 22.03.0-edge, DSL2 is the default syntax.
To minimize disruption on existing pipelines, Nextflow Tower version 22.1.x and later are configured to default Nextflow head jobs to DSL 1 for a transition period (ending TBD).
You can force your Nextflow head job to use DSL2 syntax via any of the following techniques:
- Adding
export NXF_DEFAULT_DSL=2
in the Advanced Features > Pre-run script field of Tower Launch UI. - Specifying
nextflow.enable.dsl = 2
at the top of your Nextflow workflow file. - Providing the
-dsl2
flag when invoking the Nextflow CLI (e.g.nextflow run ... -dsl2
)
Q: Can Tower to use a Nextflow workflow stored in a local git repository?
Yes. As of v22.1, Nextflow Tower Enterprise can link to workflows stored in "local" git repositories. To do so:
- Volume mount your repository folder into the Tower Enterprise
backend
container. - Update your
tower.yml
with the following configuration:1 2 3 4
tower: pipeline: allow-local-repos: - /path/to/repo
Note: This feature is not available to Tower Cloud users.
Q: Am I forced to define sensitive values in tower.env
?tower.yml
or - in the case of a Kubernetes deployment - reference data from a secrets manager like Hashicorp Vault.
Please contact Seqera Labs for more details if this is of interest.
tw CLI#
Q: Can a custom run name be specified when launch a pipeline via the tw
CLI?
Yes. As of tw
v0.6.0, this is possible. Example: tw launch --name CUSTOM_NAME ...
Workspaces#
Q: Why is my Tower-invoked pipeline trying to contact a different Workspace than the one it was launched from?
This problem will express itself with the following entry in your Nextflow log: Unexpected response for request http://YOUR_TOWER_URL/api/trace/TRACE_ID/begin?workspaceId=WORKSPACE_ID
.
This can occur due to the following reasons:
- An access token value has been hardcoded in the
tower.accessToken
block of yournextflow.config
(either via the git repository itself or override value in the launch form). - In cases where your compute environment is an HPC cluster, the credentialized user's home directory contains a stateful
nextflow.config
with a hardcoded token (e.g. `~/.nextflow/config).
Q: What privilege level is granted to a user assigned to a Workspace both as a Participant and Team member?
It is possible for a user to be concurrently assigned to a Workspace both as a named Participant and member of a Team. In such cases, Tower will grant the higher of the two privilege sets.
Example:
- If the Participant role is Launch and the Team role is Admin, the user will have Admin rights.
- If the Participant role is Admin and the Team role is Launch, the user will have Admin rights.
- If the Participant role is Launch and the Team role is Launch, the user will have Launch rights.
As a best practice, Seqera suggests using Teams as the primary vehicle for assigning rights within a Workspace and only adding named Participants when one-off privilege escalations are deemed necessary.
Amazon#
EC2 Instances#
Q: Can I run a Nextflow head job on AWS Gravitron instances?
No. Nextflow does not yet run on ARM-based compute instances.
ECS#
Q:How often are docker images pulled by the ECS Agent?
As part of the AWS Batch creation process, Tower Forge will set ECS Agent parameters in the EC2 Launch Template that is created for your cluster's EC2 instances:
- For clients using Tower Enterprise v22.01 or later:
- Any AWS Batch environment created by Tower Forge will set the ECS Agent's
ECS_IMAGE_PULL_BEHAVIOUR
set toonce
.
- Any AWS Batch environment created by Tower Forge will set the ECS Agent's
- For clients using Tower Enterprise v21.12 or earlier:
- Any AWS Batch environment created by Tower Forge will set the ECS Agent's
ECS_IMAGE_PULL_BEHAVIOUR
set todefault
.
- Any AWS Batch environment created by Tower Forge will set the ECS Agent's
Please see the AWS ECS documentation for an in-depth explanation of this difference.
Note: This behaviour cannot be changed within the Tower Application.Q: We encountered an error saying unable to parse HTTP 429 response body.
CannotPullContainerError: Error response from daemon: error parsing HTTP 429 response body: invalid character 'T' looking for beginning of value: "Too Many Requests (HAP429)"
This is because of the dockerhub rate limit of 100 anonymous pulls per 6 hours. We suggest to use the following on your launch template in order to avoid this issue:
echo ECS_IMAGE_PULL_BEHAVIOR=once >> /etc/ecs/ecs.config
Q: Help! My job failed due to a CannotInspectContainerError error.
There are multiple reasons why your pipeline could fail with an Essential container in task exited - CannotInspectContainerError: Could not transition to inspecting; timed out after waiting 30s
error.
Please try the following:
- Upgrade your ECS Agent to 1.54.1 or newer (instructions for checking your ECS Agent version);
- Provision more storage space for your EC2 instance (preferrably via ebs-autoscaling to ensure scalability).
- If the error is accompanied by
command exit status: 123
and apermissions denied
error tied to a system command, please ensure that the binary is set to be executable (i.e.chmod u+x
).
Queues#
Q: Does Nextflow Tower support the use of multiple AWS Batch queues during a single job execution?
Yes. Even though you can only create/identify a single work queue during the definition of your AWS Batch Compute Environment within Nextflow Tower, you can spread tasks across multiple queues when your job is sent to Batch for execution via your pipeline configuration.
Adding the following snippet to either your nextflow.config or the Advanced Features > Nextflow config gile field of Tower Launch UI, will cause processes to be distributed across two AWS Batch queues, depending on the assigned named.
1 2 3 4 5 6 7 8 9 10 11 12 13 |
|
Security#
Q: Can Tower connect to an RDS instance using IAM credentials instead of username/password?
No. Nextflow Tower must be supplied with a username & password to connect to its associated database.
Storage#
Q: Can I use EFS as my work directory?
As of Nextflow Tower v21.12, you can specify an Amazon Elastic File System instance as your Nextflow work directory when creating your AWS Batch Compute Environment via Tower Forge.
Q: Can I use FSX for Luster as my work directory?
As of Nextflow Tower v21.12, you can specify an Amazon FSX for Lustre instance as your Nextflow work directory when creating your AWS Batch Compute Environment via Tower Forge.
Q: How do I configure my Tower-invoked pipeline to be able to write to an S3 bucket that enforces AES256 server-side encryption?
If you need to save files to an S3 bucket protected by a bucket policy which enforces AES256 server-side encryption, additional configuration settings must be provided to the nf-launcher script which invokes the Nextflow head job:
-
Add the following configuration to the Advanced options > Nextflow config file textbox of the Launch Pipeline screen:
1 2 3 4 5
aws { client { storageEncryption = 'AES256' } }
-
Add the following configuration to the Advanced options > Pre-run script textbox of the Launch Pipeline screen:
1
export TOWER_AWS_SSE=AES256
Note: This solution requires at least Tower v21.10.4 and Nextflow 22.04.0.
Azure#
AKS#
Q: Why is Nextflow returning a "... /.git/HEAD.lock: Operation not supported" error?
This problem can occur if your Nextflow pod uses an Azure Files-type (SMB) Persistent Volume as its storage medium. By default, the jgit
library used by Nextflow attempts a filesystem link operation which is not supported by Azure Files (SMB).
To avoid this problem, please add the following code snippet in your pipeline's pre-run script field:
1 2 3 4 |
|
Batch#
Q: Why is my Azure Batch VM quota set to 0?
In order to manage capacity during the global health pandemic, Microsoft has reduced core quotas for new Batch accounts. Depending on your region and subscription type, a newly-created account may not be entitled to any VMs without first making a service request to Azure.
Please see Azure's Batch service quotas and limits page for further details.
SSL#
Q: "Problem with the SSL CA cert (path? access rights?)" error
This can occur if a tool/library in your task container requires SSL certificates to validate the identity of an external data source.
You may be able to solve the issue by:
- Mounting host certificates into the container (example).
Google#
Retry#
Q: How do I make my Nextflow pipelines more resilient to VM preemption?
Running your pipelines on preemptible VMs provides significant cost savings but increases the likelihood that a task will be interrupted before completion. It is a recommended best practice to implement a retry strategy when you encounter exit codes that are commonly related to preemption. Example:
1 2 3 4 5 |
|
Q: What are the minimum Tower Service account permissions needed for GLS and GKE?
The following roles are needed to be granted to the nextflow-service-account
.
- Cloud Life Sciences Workflows Runner
- Service Account User
- Service Usage Consumer
- Storage Object Admin
For detailed information, please refer to this guide.
Kubernetes#
Q: Pod failing with 'Invalid value: "xxx": must be less or equal to memory limit' error
This error may be encountered when you specify a value in the Head Job memory field during the creation of a Kubernetes-type Compute Environment.
If you receive an error that includes field: spec.containers[x].resources.requests
and message: Invalid value: "xxx": must be less than or equal to memory limit
, your Kubernetes cluster may be configured with system resource limits which deny the Nextflow head job's resource request. To isolate which component is causing the problem, try to launch a Pod directly on your cluster via your Kubernetes administration solution. Example:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 |
|
On-Prem HPC#
Q: "java: command not found"
When submitting jobs to your on-prem HPC (regardless of whether using SSH or Tower-Agent authentication), the following error may appear in your Nextflow logs even though you have Java on your $PATH environment variable:
1 2 3 4 |
|
Possible reasons for this error:
- The queue where the Nextflow head job runs in a different environment/node than your login node userspace.
- If your HPC cluster uses modules, the Java module may not be loaded by default.
To troubleshoot:
- Open an interactive session with the head job queue.
- Launch the Nextflow job from the interactive session.
- If you cluster used modules:
- Add
module load <your_java_module>
in the Advanced Features > Pre-run script field when creating your HPC Compute Environment within Nextflow Tower.
- Add
- If you cluster does not use modules:
- Source an environment with java and Nextflow using the Advanced Features > Pre-run script field when creating your HPC Compute Environment within Nextflow Tower.