You need to ensure that all instances of s3.docker.mender.io are replaced. It looks like it is using that name somewhere else.
You mean, we should remove the lines having “s3.docker.mender.io” from “prod.yml.template” as below screenshot?
Yes. That likely also means you need to regenerate your certificates since that name is likely embedded in them.
Has anyone solved the issue?
Deployments have not worked for us either since V2.7.0. I can upload and download artifacts and trigger deployments via the web interface without any problems. But the devices can not load them, it appears the “SerializationError” already mentioned here.
Maybe an interesting discovery. It seams the api-gateway forward the request from the device to the gui container instead of the deployments container. According to the api-gateway log the reqest is sent to docker internal ip 10.21.255.130, but this is the gui container.
xxx.xxx.xxx.xxx - - [11/Jun/2021:15:35:39 +0000] "POST /api/devices/v1/deployments/device/deployments/next HTTP/1.1" 503 76 "-" "-" 64 "deployments@docker" "http://10.21.255.130:8080" 2ms
xxx.xxx.xxx.xx - - [11/Jun/2021:15:35:40 +0000] "GET /api/devices/v1/deployments/device/deployments/next?artifact_name=my_artifact&device_type=my_device HTTP/1.1" 503 76 "-" "-" 65 "deployments@docker" "http://10.21.255.130:8080" 3ms
gui container network:
"Networks": {
"menderproduction_mender": {
"IPAMConfig": null,
"Links": null,
"Aliases": [
"mender-gui",
"80257243e507"
],
"NetworkID": "7934f1c0955f6550f3f122521673638200f09e1239eb9b0103827d2c638ee95f",
"EndpointID": "1e5b9f576a9338b89d4281ae0210c3a87dc571d0e0186ae6ad85db8c91973508",
"Gateway": "10.21.255.129",
"IPAddress": "10.21.255.130",
"IPPrefixLen": 26,
"IPv6Gateway": "",
"GlobalIPv6Address": "",
"GlobalIPv6PrefixLen": 0,
"MacAddress": "02:42:0a:15:ff:82",
"DriverOpts": null
}
deployment container network:
"Networks": {
"menderproduction_mender": {
"IPAMConfig": null,
"Links": null,
"Aliases": [
"mender-deployments",
"711cf37570be"
],
"NetworkID": "7934f1c0955f6550f3f122521673638200f09e1239eb9b0103827d2c638ee95f",
"EndpointID": "d8763abce27472628f08815aaf0bc0f3557dd80017663b472eea6d303ab65e6c",
"Gateway": "10.21.255.129",
"IPAddress": "10.21.255.143",
"IPPrefixLen": 26,
"IPv6Gateway": "",
"GlobalIPv6Address": "",
"GlobalIPv6PrefixLen": 0,
"MacAddress": "02:42:0a:15:ff:8f",
"DriverOpts": null
}
Could this be the cause, or rather a symptom?
Same thing here. At this point I’m forced to downgrade to 2.6. It’s a shame because the new features and the Web app restyling were great in 2.7.
Widening the net in case others have ideas. cc @Alan @kacf @kjaskiewiczz @lluiscampos @oleorhagen
Drew
Guys, pulling directly from master branch and reinstalling the server worked for me.
These are the differences I have in prod.yml between 2.7 tagged version and master.
You can see changes in the certificate lines, that now involves traefik. I think this can be the trick. Someone more experienced can explain the meaning of these edited lines?
@Dave and all, could you please check this by Drew: Fix broken STORAGE_PROXY_DOMAIN_NAME by drewmoseley · Pull Request #1468 · mendersoftware/mender-docs · GitHub with the above screen by @andrea.nencini (thanks) you should be able to figure it out, before we fix it.
peter
The latest version from master branch works also for me (no new installation, just upgrade).
Thank you @andrea.nencini !
hello,
this was fixed and should now be working everywhere.
peter