Connect API: /actions/run endpoint taking 6-10s for simple actions

[Afstkla]

Description

When using the Pipedream Connect API to run actions, the POST /connect/{project_id}/actions/run endpoint takes 6-10 seconds when stash_id: "" is included, but only 1.5 seconds when stash_id is omitted.

Root Cause Found

Sending stash_id: "" (empty string) to initialize a new stash causes a massive performance penalty.

Request	Response Time
Without stash_id	~1.5s
With stash_id: ""	~6-10s
With stash_id: "<actual_id>"	(not yet tested)

Environment

• API endpoint: https://api.pipedream.com/v1/connect/{project_id}/actions/run
• Action tested: trengo-list-messages

Reproduction

Fast request (~1.5s):

curl -X POST 'https://api.pipedream.com/v1/connect/{project_id}/actions/run' \
  -H 'Authorization: Bearer ...' \
  -H 'Content-Type: application/json' \
  -d '{
    "id": "trengo-list-messages",
    "configured_props": { ... },
    "external_user_id": "..."
  }'

Slow request (~6-10s):

curl -X POST 'https://api.pipedream.com/v1/connect/{project_id}/actions/run' \
  -H 'Authorization: Bearer ...' \
  -H 'Content-Type: application/json' \
  -d '{
    "id": "trengo-list-messages",
    "configured_props": { ... },
    "external_user_id": "...",
    "stash_id": ""
  }'

Impact

We use stash_id: "" on the first action of a workflow to initialize a stash for sharing data between subsequent actions. This 5-8 second penalty on stash creation significantly impacts our workflow execution times.

Expected Behavior

Creating a new stash (via stash_id: "") should not add significant latency compared to requests without stash.

Reference

The Pipedream Connect demo at https://pipedream.com/connect/demo?app=trengo&component=trengo-list-messages&tab=debug shows runtimes of 1-2 seconds, which matches our results when stash_id is omitted.