Postgres node with ssh tunnel - random Connection refused

[andrsp]

Bug Description

I'm using postgres node in workflows with option to connect via ssh tunnel.
After upgrading from 1.73 to 1.75.2 (and 1.76.0) some of workflow executions started failing on postgres node with error which looks like this:

Connection refused
127.0.0.1:44563

looks like ssh tunnel closed after some time (5m?) and reconnecting not working as expected

error log from node:

Node type: n8n-nodes-base.postgres
Node version: 2.5 (Latest)
n8n version: 1.76.0 (Self Hosted)

Stack trace

NodeOperationError: Connection refused
 at parsePostgresError (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-nodes-base@file+packages+nodes-base_@[email protected]_@aws-sdk+credential-provi_itzq4gjm3572fjrrbfktkg2f4u/node_modules/n8n-nodes-base/dist/nodes/Postgres/v2/helpers/utils.js:102:12)
 at /usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-nodes-base@file+packages+nodes-base_@[email protected]_@aws-sdk+credential-provi_itzq4gjm3572fjrrbfktkg2f4u/node_modules/n8n-nodes-base/dist/nodes/Postgres/v2/helpers/utils.js:205:31
 at processTicksAndRejections (node:internal/process/task_queues:95:5)
 at ExecuteContext.execute (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-nodes-base@file+packages+nodes-base_@[email protected]_@aws-sdk+credential-provi_itzq4gjm3572fjrrbfktkg2f4u/node_modules/n8n-nodes-base/dist/nodes/Postgres/v2/actions/database/select.operation.js:85:12)
 at ExecuteContext.router (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-nodes-base@file+packages+nodes-base_@[email protected]_@aws-sdk+credential-provi_itzq4gjm3572fjrrbfktkg2f4u/node_modules/n8n-nodes-base/dist/nodes/Postgres/v2/actions/router.js:59:26)
 at ExecuteContext.execute (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-nodes-base@file+packages+nodes-base_@[email protected]_@aws-sdk+credential-provi_itzq4gjm3572fjrrbfktkg2f4u/node_modules/n8n-nodes-base/dist/nodes/Postgres/v2/PostgresV2.node.js:16:16)
 at WorkflowExecute.runNode (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-core@[email protected][email protected][email protected]_/node_modules/n8n-core/dist/execution-engine/workflow-execute.js:632:19)
 at /usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-core@[email protected][email protected][email protected]_/node_modules/n8n-core/dist/execution-engine/workflow-execute.js:890:51
 at /usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-core@[email protected][email protected][email protected]_/node_modules/n8n-core/dist/execution-engine/workflow-execute.js:1224:20

switching to direct connection to postgres without tunnel fixes problem

To Reproduce

1. set up postgres credentials with ssh tunnel
2. set up workflow with postgres node using these credentials
3. execute flow continuously for 5-10 minutes
4. after ~5min you will start see errors

Expected behavior

postgres node works same way with or without ssh tunnel

Operating System

ubuntu 22.04 / docker

n8n Version

1.76

Node.js Version

2.5

Database

PostgreSQL

Execution mode

queue

[Joffcom]

Hey @andrsp,

We have created an internal ticket to look into this which we will be tracking as "N8N-8213"