currently only flushes changes some time after chunk unload, which often only happens on server shutdown, causing loss of hours of data when the server crashes (as happened a few times already)
flushing every 30min/5min/whatever would make crashes less painful, and possibly also shorten shutdown/restart times
shadowtrot said:
it would be better to save async to cut main thread load, but for that it needs updated to be safe to do that async
currently it synchronizes on the input and output queues and the chunkcoord when unloading to make sure main thread doesn't interfere
currently only flushes changes some time after chunk unload, which often only happens on server shutdown, causing loss of hours of data when the server crashes (as happened a few times already)
flushing every 30min/5min/whatever would make crashes less painful, and possibly also shorten shutdown/restart times
shadowtrot said: