As subject states: disk ran full at 0% free space. Disk was expanded and graylog-server restarted.
Looking in System > Nodes, I can see that after graylog-server restart the buffers are working like expected for a few seconds, maybe half a minute, appending and reading thousands of messages pr second. THen it changes to 0 messages or perhaps some random small number, same for both buffers, but the unprocessed messages number grows steadily.
I have wiped the journal folder, restarted both graylog-server and the VM multiple times.