Support

This looks like a timeout trying to upload a 20Mb chunk of data to Dropbox.

Go to wp-admin, Akeeba Backup, Configuration, find the Post-processing engine row, click on the Configure button next to it. Set the Chunk size to 5 and retry. If you get a problem still please check with your host if the PHP cURL extension is enabled for PHP 8.0.25 on your site. If it is, please send me the new backup log file.

I was afraid I would see that. This looks like a problem on your hosting side. Your host either has the slowest uplink connection I've seen in well over 20 years or has a misbehaving caching proxy for outbound connections. Here's how I can tell:

DEBUG |20230115 11:48:31|Akeeba\Engine\Postproc\Dropbox2::processPartChunked - New upload session pid_upload_session:ABIF34EdrkGNM-i3wxw97V3oP-fkA_QPoIw4c0lsN7Osv11A
DEBUG |20230115 11:48:31|Akeeba\Engine\Postproc\Dropbox2::processPartChunked - Uploading chunked part
DEBUG |20230115 11:54:21|Kettenrad :: Attempting to load from database (backend)

This tells me that Akeeba Backup started to upload 5MiB of data at 11:48:31 and the host killed that process at 11:54:11. The next ten seconds to 11:54:21 is the countdown in the interface to retry the backup. Four and a half minutes trying and failing to upload 5MiB would make for a transfer rate less than 150Kbps which is absurd — we were using cheap servers with 10Mbps uplinks 20 years ago.

The Dropbox API does not let you use a chunk size smaller than 5MiB to avoid that timeout coming from your server.

You really need to talk to your host. It's very likely they have a misbehaving proxy for outbound connections which is blocking the transfer, making it appear on the PHP side as though it's taking forever.

It is very likely that it's a problem with just this one server (physical or virtual). It would certainly not be the first time we see that.

I've  had two of my own sites on the same host, same physical server, different virtual server, where one would work fine and the other would fail uploading to AWS every few weeks. It turns out the deployment script had bombed out when setting up ntpd (the time synchronisation server). Every few weeks the clock would drift far enough that we went past the "wiggle room" allowed by AWS. Rebooting the virtual server fixed the problem because ntpd would sync the time on reboot — but not periodically while the server was up. It took a third level systems engineer to spot the problem and fix it. I had no such problem with that host on any other site before or since.