add expected 'df' output to README about irods_client_nfsrods

[michael-conway]

What's the best approach to giving a meaningful number? Or is this about setting a non-zero value?

…

On Wed, Aug 21, 2019 at 9:56 AM Terrell Russell ***@***.***> wrote: NFSRODS populates the available and used space fields as 0. This signals to a default df request to ignore the mountpoint information for NFSRODS. Live NFSRODS mountpoint information can be viewed via df -a. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#25?email_source=notifications&email_token=AAIL4LJGSBZJGJ2KCBKSMTTQFVCSDA5CNFSM4IOHKDAKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HGQVX5Q>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAIL4LP2XCMRKOTPQUDQ5ZTQFVCSDANCNFSM4IOHKDAA> .

[trel]

We've settled on reporting zeroes for now since anything else would be misleading.

• There is no 'actual' number (iRODS has multiple resources, possibly including AWS S3...).
• Arbitrary 'ReallyBigNumber' is misleading and could cause certain problems to other programs using that information to make decisions.
• Zero is a known pattern for other 'silent' mountpoints.

If there are good ideas, please share - otherwise, we're going with 0 for now.

[anderbubble]

We're working on deploying quota support as seen here:

https://github.com/irods/irods_policy_examples/blob/master/irods_policy_logical_quotas.re

I'd love to have these quotas reflected in the df output. Maybe there should just be a standard set of irods metadata that should get reported out via NFSRODS? Or maybe it should read df data out of a PEP? (Sorry if these are the wrong terms; let me know if this isn't clear.)

[michael-conway]

It would not be terribly difficult to service avus or add api support for quotas gen there was a definite standard way in irods? I did t know what the current state of quotas was.

…

On Tue, Sep 17, 2019, 2:40 PM Jonathon Anderson ***@***.***> wrote: We're working on deploying quota support as seen here: https://github.com/irods/irods_policy_examples/blob/master/irods_policy_logical_quotas.re I'd love to have these quotas reflected in the df output. Maybe there should just be a standard set of irods metadata that should get reported out via NFSRODS? Or maybe it should read df data out of a PEP? (Sorry if these are the wrong terms; let me know if this isn't clear.) — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#25?email_source=notifications&email_token=AAIL4LPTODM7YPVOCX4DSJTQKEQB3A5CNFSM4IOHKDAKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD65P7PY#issuecomment-532348863>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAIL4LLIX3HORCOAWNFHBX3QKEQB3ANCNFSM4IOHKDAA> .

[anderbubble]

To be fair, I think the current state is "nonexistent" except that we asked for it and that policy is what came out. :) We haven't deployed or tested it yet; but that's basically my next step for our environment (in competition with "get NFSRODS to work").

[trel]

The linked logical quotas policy is for 'all users' under a particular logical path. It handles keeping track of totals in the system via pre- and post- PEPs when data is added/removed from the system.
They possibly could be represented to df in the way you're describing - however...

iRODS also has the concept of both total and per-resource quotas as defined per-user and per-group. I can't figure out how to represent those to df at this time - or whether they get in the way at all.

Also, NFSRODS is only a single iRODS client - so we'd need to think through what representing 'capacity' means for other clients and how they can be fairly informed of the same type of information.