I’m doing some disk storage benchmarking and need to know the on-disk footprint for various use cases.
Currently, I specify write-ahead and posting lists directories via the “-p” and “-w”, setting both to the same directory. Then I run some mutations, wait for the batches to process, and then get the size of that directory.
Don’t do that. These directories need to be separate.
Dgraph stores it’s state in p directory, and write ahead log in w. If you want to check the on-disk files, you should write your data, and shutdown dgraph, so it will push everything it has in memory to disk. Note that the disk usage shown by w can vary depending upon how many logs it carries. These logs though get periodically deleted, and will get purged later.