Repository navigation
Connect Timeout Error when parsing NMR data from URL #137
Description
Describe the bug
The /spectra/parse/url API endpoint fails with a Connect Timeout Error when attempting to fetch and parse NMR data from a URL.
To Reproduce
Steps to reproduce the behavior:
- Make a POST request to
/spectra/parse/url - Use the following request body:
{
"auto_detection": true,
"auto_processing": true,
"capture_snapshot": true,
"url": "https://s3.uni-jena.de/nmrxiv/production/archive/1368b8fd-3af7-48f7-8b6d-13de9c1ecf8b/492-62-6.zip"
}- Observe the error response
Error Details
The API returns a detailed error message:
NMR CLI error: node:internal/deps/undici/undici:14902
Error.captureStackTrace(err);
^
TypeError: fetch failed
at node:internal/deps/undici/undici:14902:13
at process.processTicksAndRejections (node:internal/process/task_queues:105:5)
at async Object.arrayBuffer (file:///app/node_modules/file-collection/lib/append/sourceItemToExtendedSourceItem.js:35:26)
at async fileItemUnzip (file:///app/node_modules/file-collection/lib/utilities/expand/fileItemUnzip.js:26:20)
at async expandAndFilter (file:///app/node_modules/file-collection/lib/utilities/expand/expandAndFilter.js:17:23)
at async FileCollection.appendExtendedSource (file:///app/node_modules/file-collection/lib/FileCollection.js:62:23)
at async Promise.all (index 0)
at async appendSource (file:///app/node_modules/file-collection/lib/append/appendSource.js:23:5)
at async FileCollection.appendSource (file:///app/node_modules/file-collection/lib/FileCollection.js:133:9)
at async FileCollection.fromSource (file:///app/node_modules/file-collection/lib/FileCollection.js:229:9)
Root Cause:
ConnectTimeoutError: Connect Timeout Error (attempted address: s3.uni-jena.de:443, timeout: 10000ms)
at onConnectTimeout (node:internal/deps/undici/undici:2718:28)
at Immediate._onImmediate (node:internal/deps/undici/undici:2699:11)
at process.processImmediate (node:internal/timers:485:21)
Activity
hamed-musallam commentedon Sep 1, 2026
The sample provided in the ticket contains 4 1D spectra (2 FID, 2 FT) and 8 2D spectra (half of which are FT). As you can imagine, automated range and zone detection across these datasets requires a significant amount of memory, which is leading to a memory leak.
This is not an issue during standard auto-processing or snapshot capturing. The issue occurs during the auto-detection step, which is where we need to focus our performance improvements. To address this, we need to investigate further how the 2D zones are detected in the processed spectra by NMRium.