pNFS server Plan B

[Willem Jan Withagen]

Hi Jordan,

To rip just a bit of your text out of context:
On 18-6-2016 22:50, Jordan Hubbard wrote:
> Some, if not most, of them are also far
> better supported under Linux than FreeBSD (I don’t think we even have
> a working ceph port yet).   

In the spare time I have left, I'm trying to get a lot of small fixes
into the ceph tree to get it actually compiling, testing, and running on
FreeBSD. But Ceph is a lot of code, and since a lot of people are
working on it, the number of code changes are big. And just keeping up
with that is sometimes hard. More and more Linux-isms are dropped into
the the code. So progress is slow. I only because it is hard to get
people to look at the commits and get them.
Current state is that I have it compile everything, and I can run 120 of
129 test with success. I once had them complete all, but then a busload
of changes were dropped in the tree. And so I needed to "start
"repairing" again.

I gave a small presentation of my work thus far at Ceph Day Cern in
Geneva. https://indico.cern.ch/event/542464/contributions/2202309/
Differences in code are not really that big in the CC-code, most of the
things to fix are additional tools that have to deal with the
infrastructure that fully assumes it is running a Linux-distro.

Next to that is Ceph going to its own diskstore system: BlueStore, where
as I hope(d) to base it on a ZFS underlying layer...
To run BlueStore AIO is needed for diskdevices, but the current AIO is
not call for call compatible, and requires a glue layer. I have not
looked into the size of the semantic problems between Linux and FreeBSD
here.

On the other hand they just declared CephFS (a posix filesystem running
on Ceph) stable and to be used.

--WjW