We’re running into an issue with uploading images via the Portainer Agent 1.23.2.
Portainer “standlone” on the same VMs works fine, but when we swap it out to the Agent version in a Docker compose we’re seeing consistent “net::ERR_CONNECTION_RESET”.
- Happens regardless of browser (fairly recent versions FireFox, Chrome, Brave and Edge all tested).
- Happens regardless of host OS (RedHat 7.7, Ubuntu 19.10, Windows 10 Pro).
- Fails using the agent on the same VM for uploads where the non-agent (standalone) Portainer has been hosted
- Works not using the agent on the same VM for uploads
- Works using the agent on the same machine uploading from (Windows 10 Pro)
- Fails on RedHat VMs - has the firewall enabled with the following ports open:
- Fails on Ubuntu Server (PC installed with Ubuntu 19.10 Desktop edition) - has no firewall running to rule it out
- All remote hosts using the agent fail anywhere after around 40 seconds with net::ERR_CONNECTION_RESET
Everything else in the Agent appears to work as expected - container, stack & swarm management all working just like the “standalone” version.
The only difference I see is a /1 (in standalone) versus a /2 (in agent) for the load API call.
http://host:9000/api/endpoints/1/docker/images/load Non-Agent based upload (Works) http://host:9000/api/endpoints/2/docker/images/load Agent based upload (Fails, except in the local only case)
Windows 10 Pro with local Docker swarm hosting and the TAR image on the same machine seemed to work with a 152MB file - taking roughly 4 seconds.
I’m beginning to believe this is a timeout issue in the Agent configuration - seeing that the agent worked locally, but failed on remote transfers.
I’m working from home these days, so testing from in the office may differ. Right now I’m uploading around 4Mbps from home.
Any help would be appreciated as we’re back to manual Docker imports until this is resolved.