[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[nfsv4] [FedFS] Meeting Minutes, 10/01/2009
FedFS Meeting Minutes, 10/01/2009
---------------------------------
Attendees
---------
Craig Everhart (NetApp)
Sorin Faibish (EMC)
Paul Lemahieu (EMC)
James Lentini (NetApp)
Pavan Mettu (Sun)
Rob Thurlow (Sun)
Minutes
-------
+ IETF Note Well Agreement
This is a reminder that our discussions are governed by the
IETF Note Well Agreement. See:
http://www.ietf.org/NOTEWELL.html
We will start each week's meeting with this announcement.
+ Review Action Items
- Rob Thurlow will consider the use of alternative LDAP ports
for an NSDB and report back with a recommendation to the group.
Close, see below.
- Trond Myklebust will consult with Craig and send
zero-configuration DNS SRV proposal
No update.
+ Austin NFS Bake-a-thon
- status of referral support in client implementations
Paul asked what the status was for referral support in client
implementations.
Rob and Pavan will be bringing a Solaris client that handles referrals
to bake-a-thon. It handles a strict referral case where the client has not
seen the filesystem before. A migration event is not supported.
A referral on mount (where the client receives a referral while
following a referral) is also not supported yet. These capabilities are
only available in the development build. Currently, the client follows the
first FSL in a referral.
Craig pointed out that although migration and replication were specified
first, implementations are adding support for referrals (namespace) first.
At the bake-a-thon, we hope to have the implementations querying the
NSDB to map FSN to FSLs.
James will be bringing an OpenLDAP NSDB and OpenDS NSDB.
- No Meeting next week
We decided to cancel next weeks meeting since many participants would be
at the bake-a-thon.
+ Requirements Draft Update
James presented the changes published today in -04. A -05 is already in
the works that will include the improvements suggested by Nico Williams's
review for the security directorate. The review contained 5 suggestions:
improvements to security considerations:
- noting the authorization separation inherent in the design
- the resource consumption considerations for NSDB objects
- implications regarding NSDB authentication
improvements to general description:
- regarding the absence of junctions in the NSDB
- the benefits of using bounded size FSNs
+ Discuss Open Issues
- Configurable NSDB port?
We reviewed the email discussions on the mailing list from this morning.
Given that Nico felt firewalls were a minor issue:
http://www.ietf.org/mail-archive/web/nfsv4/current/msg07440.html
and nobody argued for requiring NSDBs to run on the standard LDAP port
we are going to consider this issue closed.
The drafts will be updated to clarify this point and note the
potential interactions with firewalls.
- FSN format: <NSDB location, LDAP search base DN, UUID>
Rob suggested that the LDAP search base DN should be an optional
field with a default value specified if it is omitted.
We discussed the advantages of keeping the FSN size bounded.
We reviewed the sizes of the various fields. The NSDB domain name
is at most 255 bytes (Section 3.1 RFC 1034 255) and the UUID is
16 bytes.
For the LDAP search base DN we discussed requiring fileservers
to support a minimum size and allowing them to support a larger
size.
256 bytes seemed like a reasonable minimum size to require fileservers
to support for the LDAP search base DN.
We also discussed specifying a maximum size for the entire FSN.
1 KB was proposed.
The consensus was that the additional flexibility allowed by
this change would be useful and therefore worth including.
- NSDB location format: hostname or SRV RR?
Nothing new to add.
James needs to get an answer to his question:
http://www.ietf.org/mail-archive/web/nfsv4/current/msg07398.html
- fedfsNfsPathname format
Nothing new to add.