Could you bypass cache when try to load node which had errors?

Some node which had errors load only data with old cache. This makes some problems when I want to reload after I fixed them.

  • Ex: I reload workflow had error, and it load old data from mysql node.

I’m not sure about this cache, but it is like a cache.
I think cache is not necessary when we reload some workflow had errors. In fact, I usually need to load new data when try to reload workflow which had errors.

Hey @cmdntd987, not sure I fully understand the problem I am afraid.

Are you saying you have updated a workflow but n8n would load an old version of that workflow when opening it?

Oh, I’m sorry because it is not clear.

I mean I have a workflow and a database to load from mysql node:

workflow (and database) → execution with errors (because of mysql database)

When I update new database, to fix errors (just edit some fields). It has problem like this:

workflow (and new database) → execution with errors (still load old database. It is not load new database which is updated)

It looks caused by cache. I do not know how to delete cache.
But I think this cache is not necessary in this case (for reload error executions)

Hm, n8n doesn’t cache database queries on its own. Is there a chance you have some additional component sitting between your database and n8n?

I have it go through Cloudflare proxies on CNAME
Does it affect to data cache?

Hey @cmdntd987, I don’t know I am afraid. This might be a question for Cloudflare.

I tried to disable Cloudflare proxy, it still have the same issue, with more nodes (like function node). I think there is something force them to load old data, instead of new one.

  • I use docker through a proxy, and not sure if it caused this

I am honestly still confused. Can you please share how the URL looks like when this is happening? Does it contain /workflow or /execution?

I see it in /execution/123
When I use re-run some error workflow like this: