IMAP Read stops working - no error/nothing/just doesn't read emails anymore

Questions
#1

Hey!

I’m a heavy IMAP Read Trigger user but I’m running into a stability problem on two seperate server instances. The IMAP Read Trigger stops working after round about 20 hours. This occures on both instances, so I don’t think it’s related to the server setup. I’m also using seperate email providers.

I can see a last run in n8n which worked out just fine and after this, nothing happens anymore until I restart the workflow. Even restarting the docker container does not help ervery time… (I tried to make a quick fix and restart the container during night via cronjob).

Are there more logs to read than the normal docker logs so I can have a look into this? The problem is: I get kind of a jam in the mailbox and n8n only proccesses the first new email - all other mails get kind a lost.

Would be great if I/we can find a solution for this or someone can advice me to find some more logs, to find out what kills the node.

#2

It seems, that there are also other people who are running into this bug.

Email Read IMAP Node works sometimes - Questions - n8n

Because I can not find any other logs and it seems that nobody knows if there are other logs I’m wondering if it is a mail provider related problem. If that’s true, perhaps anybody can recommend a working mail provider. Sounds a little bit unrealistic to me that it’s not working with gmail but with another one but you never know :slight_smile:

#3

The IMAP Read node usually stops working here after an update of n8n. What fixes the issue for me is deactivating and reactivating the workflow. After that, the workflow keeps working for longer times.

The failure after an update of n8n (dockerized) is consistent.

#4

Hi all. I recently started using an IMAP trigger node and started facing this issue as well.

IMAP trigger node containing workflow needs to be deactivated and reactivated to get the workflow working again.

Any advise on this, @jan @maxT ?

Thanks.

#5

Currently sadly not but we have already assign a developer to do some digging.