With this Mender server, from time to time I am not able to upload the mender artifacts successfully. Uploading artifacts results in the following error message:
Strangely after restarting the server by calling ./run down and ./run up -d, I am able to upload the same artifcat successfully. I would like to fix this issue permanently. In the previous mender versions 1.3 and 1.7, I am not facing this issue. Any info on the topic would be helpful
I believe this is because of a gateway timeout bug. This has been fixed in 2.4.0, which is scheduled to be released very soon. I encourage you to try it as soon as it’s out. Note that the 2.4.0b1 beta does not contain the fix, it was fixed after that.
I’m suffering a similar issue, but all the time, while uploading large (1 GB and up) rootfs artifacts to 2.3.
The UI shows a upload bar that goes to 100%, stays there a while and then disappears, but the artifact file doesn’t show up in Releases. I’ve looked at the logging of the server and it seems that minio receives a PUT command for the file and then after a few minutes a DELETE command for it. But I haven’t seen those for all my attempts to upload, so I wasn’t sure that that was related.
Is this caused by the same bug? Because I’ve been trying to determine the cause for this for the last few days and it was driving me insane. Small module artifacts would load correctly (but for some reason can’t be deployed per device, only per group, which might be another bug)
If this is fixed in version 2.4, can we update the server but leave our client devices at 2.3? Because we just got those working and we don’t want to start another upgrade cycle so soon.
Yes, it’s the same bug, and leaving clients at the previous version is fine. In fact, upgrading the server before the clients is the recommended method.
Ok, thanks for the quick response. Is there a document or web page that describes the steps to follow to do an update of a running Mender Server?
With running I mean a server that is already configured.
— edit —
Never mind, I just saved my current production dir and used the 2.4.0 GIT branch to install a new instance and then copied the relevant information from the old production dir (making sure that I created a new prod.yml using the information of the old one and not straigh copying the file).
