I add a really huge directory to ipfs with command:
ipfs add --nocopy -r /root/.ipfs/my_super_huge_mounted_folder
Then I FORGET the hash of that folder.
(How to get access to the hash of that folder?)
ipfs filestore ls
gives some hashes which I believe are the hashes of segments of each file, but nowhere I found a hash of a directory.
Put it in mfs after adding
ipfs files cp /ipfs/<cid> /the-thing-I-just-added .
You can always see the hash with
ipfs files stat /the-thing-I-just-added.
You could also potentially find out the older hash with
ipfs pin ls --type=recursive.
I was kind of curious what the overhead of putting something into MFS is.
I don’t think there is much overhead. MFS just manages references in a DAG, it does not copy blocks around. Of course an update in a leave on a 1000-level deep MFS DAG may lag a bit.
I am not so advanced user. Can you help a little bit more?
Do you propose to drop filestore for mfs?
Is there a guarantee that my files will not duplicated (filestore guarantees that).
I forget the hash of root folder added in IPFS. So what is that command which you propose to me?
ipfs files cp /ipfs/<cid> /the-thing-I-just-added
My 10TB data are in folder:
what you prefer.
So, how can I use mfs without duplication of data?
No, you need to put it on MFS after you do your regular
add. That way you can give a name to what you added (and modify the content as needed). See
ipfs files --help for info about MFS in general.
So If I forget the root folder
Qm... I must do the following:
- Stop ipfs daemon
- Delete /root/.ipfs folder
ipfs config --json Experimental.FilestoreEnabled true
ipfs daemon &
ipfs add --nocopy -r /root/.ipfs/nas/folder1/shared_folder and get the e.g.
ipfs files cp ipfs/Qm1234567...... /shared_folder
Now I cannot forget the root folder’s
Qm... because I always can run:
ipfs files stat /shared_folder
which gives me the
Qm... of root folder.
Also I can run:
ipfs files stat /shared_folder/a_file_or_folder
to get the corresponding