Unable to sign without access token - OAuth Google

[Anakin_Skywalker]

Hey people!

I’m working on automations and AI agents using Google Suite apps with n8n. I started on the hosted n8n plan and then moved to running it on a VPS with a server setup. However, I’ve hit a problem since making the switch.

I can connect to Gmail, Calendar, etc., but when executing steps, I get this error:
“Unable to sign without access token.”

Here’s my setup:

• I have the redirect URL, testing user, and authorized domain configured.
• The Gmail app is enabled, and I’m connecting with the client ID and secret key.
• Everything works fine during the connection step, but execution fails.

It worked perfectly on both n8n hosted and when I hosted it on my PC. My domain has SSL, and I can’t figure out what’s causing the issue.

Some context:

• My Google app is still in “testing” mode.
• I added myself (the same credentials I’m connecting with) as a testing user.

I’m not very technical, so any guidance or ideas would be greatly appreciated! Thanks in advance for your help!

[FredericS]

Hi @Anakin_Skywalker,
I got the exact same issue with Google Analytics! Account is connected but when using the GET report, got the same error message “Unable to sign without access token”.

And I am using Gmail with the same Google account, no error but, the difference is I just created the new credential for GA so I am wondering if it is linked with the last update of n8n?
Account is properly connected:

Not sure what else to try…

[Geoffrey_T]

I’ve been having the same issue with pretty much all of my Oauth 2 api connections since updating to 1.74.1. Everything worked fine in the previous version.

[jordi]

It also happens to me, with other local installations with older versions it works perfectly, but with 1.74.1, all connections to Google services appear with the message: “unable to sign without access token”.

[Geoffrey_T]

I’ve managed to get my Oauth 2 workflows working again, by removing proxying for my n8n domain on Cloudflare. I then set N8N_PROXY_HOPS to 1 in the envirnoment settings and re-enabled the Cloudflare proxy and everything appears to be working fine with 1.74.1

[Anakin_Skywalker]

@Geoffrey_T

emoving proxying

thank you for sharing!
have you re-connected or did anything else on top of those? It’s not working for me.

Server Avatar:

Turned Off and On Proxy on CloudFlare and retested the steps. still nothing…

[Anakin_Skywalker]

UPDATE: It works now! Thank you

[jordi]

I have self-hosted n8n on a PC with Docker-Desktop with the self-hosted package with Qdrant, ollama, etc… Does this solution also work? The problem with updating the self-hosted package is that it always installs the latest version of n8n, it does not allow you to choose the “latest”, stable. Or at least I don’t know how to do it.

Thanks

[FredericS]

Thanks @Geoffrey_T - I am using DigitalOcean and I added the rule, no luck. And I am also using Cloudflare and proxy is off, any other steps that I may miss? Or anyone is using DO and solve the issue?

[Geoffrey_T]

For image in your docker compose, you can specify the version number, for eg: docker.n8n.io/n8nio/n8n:1.74.0

if you just use the default docker.n8n.io/n8nio/n8n it’ll use the latest version.

[jordi]

Thank you very much for answering, precisely when I managed to install the desired version, I saw that you have answered me with the solution, I just wanted to comment that from what I read in the documentation, the option n8n:latest, always installs the “stable” version or, in this case, the 1. 73.1, and at least in installing the package self-hosted-ai-starter-kit, it is not, with this option installs the latest non-stable, ie 1.74.1. 73.1, and at least in the installation of the self-hosted-ai-starter-kit package, it is not so, with this option installs the latest not stable, ie 1.74.1. Until I have not indicated n8n:1.73.1, it has not installed the stable version or as you indicate, the desired version. I attach images of docker-compose.yml. I understand that in future versions will fix it, will be attentive from now on not to install anything that is not “stable”.

Regards

[FredericS]

Thanks! I forced n8n to 1.73.1 but still no luck…
I saw an update today for version 1.7.5: Don’t break oauth credentials when updating them and allow fixing broken oauth credentials by repeating the authorization flow
I will try that!

UDPATE: Nope! 1.75.0 didn’t solve my issue… I am giving up for now! Thank you all for the help. I learned a lot, really appreciate it.

[Ramon_Tomzer]

updated to 1.75.0 and added the line about proxy hops to the docker compose file and reauthed in the workflow I was having issues with and it’s working now. Thank you @Geoffrey_T

[Ken_Ssoon]

Same here…2 days ago it was working fine. Yesterday, it stopped working and this morning I upgraded to the latest version and still not working. I’m pretty sure this a n8n issue, because I have the same workflow using make.com and it works without issues.
Mine is self-hosted on repocloud.io. Anyone having the same issue on the n8n cloud?

[Macflight38]

Ive the same problem on calendar but not on gmail … Ive restart my server VPS added N8N_PROXY_HOPS=1 but still dont work …on 1.74.2

So my solution was to erase the credential and recreate it for me. and now it work

[NoelAPH]

Thank you for posting this. I had the same problem and was looking at some type of token issue on the google end. The proxy hops env var fixed it for me as well.

[jere_gomez]

I was having the same issue some days ago. Version 1.74.1 installed locally via npm.
It was solved by updating to 1.75.2.

[Robm]

Deleting and recreating the credential did the trick

[freddy-schuetz]

Thanks. Deleting credentials and create new worked fine!

[Ken_Ssoon]

What version are you running? I deleted everything from the workflow credentials & the API credentials, even tried creating a new workflow and I’m getting the same result.