Skip to content

Troubleshooting

Use this page when a pipeline cannot be published or a run fails.

Pipeline will not publish

Required input is disconnected

A block has an input with no upstream connection.

Example:

Python:input

Connect a compatible source/output.

Port types do not match

Connections must use the same type.

Invalid:

Arrow table → JSON

Valid:

Arrow table → Arrow table

Input already has a connection

Each input accepts one upstream connection.

If you need two datasets, create two named inputs instead.

Pipeline contains a loop

Pipelines cannot feed data back into an earlier block.

Remove the connection that creates the cycle.

Python or R block fails

Input name is wrong

If the block input is:

measurements

the script must use:

ft.input_table("measurements")

not:

ft.input_table("input")

unless the port is actually named input.

Column does not exist

Inspect the incoming data and check the exact column names.

Python example:

print(df.columns)

Then use the run output/error details to confirm what the script received.

Script timed out

The block exceeded its configured timeout.

Possible fixes:

  • filter the source data more aggressively;
  • simplify the calculation;
  • split the transformation across blocks;
  • increase the timeout when appropriate.

Script used too much memory

Large Pandas operations can use substantial memory.

Consider:

  • selecting fewer columns;
  • filtering rows in the source query;
  • aggregating in Flight SQL first;
  • avoiding unnecessary dataframe copies.

Package is unavailable

Pipeline scripts cannot install packages from the internet while running.

Contact your FlaskTrack administrator if an additional Python or R package needs to be added to the pipeline runtime.

File problems

FlaskTrack file cannot be read

Check that:

  • the file still exists;
  • the file belongs to your organization;
  • the correct file was selected as the source;
  • your script is using the correct input alias.

Generated file is rejected

Files created by a script must be written into the pipeline output workspace.

Python example:

path = "/workspace/output/result.csv"
df.to_csv(path, index=False)

ft.output_file(
    "result",
    path,
    content_type="text/csv",
)

The output alias must also exist on the block.

Output block fails

Check:

  • the incoming connection;
  • the input type;
  • destination path;
  • filename.

Example destination:

analysis/qc

Do not use:

/analysis/qc

or paths containing:

..

Run remains queued

A queued run is waiting for a Data Pipeline worker.

If runs remain queued for an unusual amount of time, contact your FlaskTrack administrator.

Run was cancelled

A cancellation request stops future pipeline progress.

Blocks that had already completed remain recorded in the run.

Start a new run if you want to execute the pipeline again.

Run failed after previously working

Check whether:

  • the source data changed;
  • a source file was replaced or removed;
  • a saved report changed;
  • the published version is the one you expected;
  • the data now contains values the script did not previously encounter.

The run page preserves the specific version and execution history to help compare successful and failed runs.