What is a reasonable value for annex.bloomcapacity
in this situation and in what unit?
local annex keys: 670671
local annex size: 2.62 terabytes
annexed files in working tree: 1410199
size of annexed files in working tree: 3.52 terabytes
combined annex size of all repositories: 2.63 terabytes
annex sizes of repositories:
2.62 TB: b2e77041-584e-4699-947d-ef5004273901 -- pudong [here]
7.39 GB: 42a53a13-7ad6-43a5-95ef-813d4f8c1a6f -- [batam_payload]
backend usage:
SHA256E: 1410199
bloom filter size: 32 mebibytes (appears too small for this repository; adjust annex.bloomcapacity)
I don't really understand what happens when the bloom filter is too small.
What will happen with too small a bloom filter is
git-annex unused
may think that some keys are used which are really not. Andgit-annex sync --content
may operate on some keys that are not in the work tree.The
git-annex info
command displays how much memory the configured bloom filters use, which is why it's reporting 32 membibytes. But the annex.bloomcapacity setting is the number of expected files in the work tree, by default 500000.It would probably make sense for you to set it to 2000000 or so unless your system has an unusually small amount of RAM.