I am running a self hosted instance of n8n. I have a worflow which accesses WooCommerce API, then narrows data for import and then is supposed to import data into Supabase. Everything runs just fine for few first lines but then, when Recursive character text splitter nears to at about 1100 counter everything slows down and becomes unsuable.
I have paid Supabase account (PRO), paid OpenAi account (Tier 3 enabled and funded) and funded OpenRouter account. I have tried everything in my power, went to many dead ends using AI and now I am desperately asking fow your help.
What can be the bottleneck, please? What can cause this massive slow down?
If there is anything I cen clear up, please ask. I am desperate…
Hi Mirek! Based on your workflow and the slowdown around 1100 items, this looks like a classic database connection/performance bottleneck. Here are the key areas to check:
**Most likely causes:**
• **Connection pooling issues** - Your Supabase instance may be hitting connection limits. Check your [supabase.com](Performance Tuning | Supabase Docs) dashboard for connection usage and consider using fewer concurrent connections in n8n
• **Batch size problems** - Processing 1100+ items individually is inefficient. Try batching your Supabase inserts (50-100 records per request) instead of single inserts
• **Resource constraints** - Your Supabase Pro plan may need more compute power for this workload
**Quick fixes to try:**
• Enable connection pooling in your Supabase settings
• Add a “Batch” node before your Supabase insert to group records
• Check your Supabase dashboard for CPU/memory usage during the slowdown
The OpenAI embeddings processing before Supabase could also be creating a bottleneck - try adding delays between API calls or reducing batch sizes there too.
