nfs-utils-2.3.1

[bdubbs@…]

New minor version.

Taking this, to see if removes the spew in syslog from nfs3 mounts - with 2.1.1 every such mount reports

NFS: bad mount option value specified: minorversion=1

[ken@…]

Oh well, it was worth a try - seems to work, but still puts that out on every nfs mount. Deferred for further investigation.

[bdubbs@…]

Replying to bdubbs@…:

Taking this, to see if removes the spew in syslog from nfs3 mounts - with 2.1.1 every such mount reports

NFS: bad mount option value specified: minorversion=1

I've seen that aslo, but fixed it with a change to fstab:

lfs78:/srv/src /usr/src nfs noauto,rw,nfsvers=3 0 0

[ken@…]

Replying to bdubbs@…:

I've seen that aslo, but fixed it with a change to fstab:

lfs78:/srv/src /usr/src nfs noauto,rw,nfsvers=3 0 0

Thanks, I thought I had tried that but I'll give it another go in the fresh build I'm about to start.

[ken@…]

Replying to ken@…:

Replying to bdubbs@…:

I've seen that aslo, but fixed it with a change to fstab:

lfs78:/srv/src /usr/src nfs noauto,rw,nfsvers=3 0 0

Thanks, I thought I had tried that but I'll give it another go in the fresh build I'm about to start.

Works nicely, thanks again. I don't have any notes on what I tried, so I assume I made a typo.

[ken@…]

The package *works* fine, but I can't get the tests to run:

make[4]: *** No rule to make target '../../support/nfs/libnfs.a', needed by 'nsm_client'.  Stop.

The only relevant rule seems to be for libnfs.la in support/nfs/Makefile and that *did* get created.

Unless anybody has any ideas, I'm minded to mark the testsuite as b0rken.

[bdubbs@…]

I found libnfs.a in support/nfs/.libs/libnfs.a

Copying that to support/nfs makes some progress, but building statdb_dump fails with undefined references to generic_make_pathname and generic_setup_basedir. Those appear to be in support/nsm/libnsm.a, but using nm, the symbols are undefined.

I agree that for now, the tests should be marked broken until upstream fixes them. What we us in other packages is:

"This packages does not come with a working test suite."

[ken@…]

Replying to bdubbs@…:

I found libnfs.a in support/nfs/.libs/libnfs.a

Copying that to support/nfs makes some progress, but building statdb_dump fails with undefined references to generic_make_pathname and generic_setup_basedir. Those appear to be in support/nsm/libnsm.a, but using nm, the symbols are undefined.

Thanks for that analysis.

I agree that for now, the tests should be marked broken until upstream fixes them. What we us in other packages is:

"This packages does not come with a working test suite."

Yes, that is what I intend to put in the book, my earlier comment was just intended as shorthand for that.

However, at the moment I have a number of issues with a different (-beta) package which will probably stop me finishing this today.

[ken@…]

Just a note that the systemd book has a note about nfsver=3, but (on sysv) that is partly out of date - with this version of nfs-utils, and also 2.2, mount works (as noted near the start), but generates a log message on every mount [ I write my backups to nfs, so I do a lot of mounting ].

[ken@…]

Replying to ken@…:

Just a note that the systemd book has a note about nfsver=3, but (on sysv) that is partly out of date - with this version of nfs-utils, and also 2.2, mount works (as noted near the start), but generates a log message on every mount [ I write my backups to nfs, so I do a lot of mounting ].

OK, so it's a long page, particularly when I expand the text to a readable size, and I'm working on a short monitor. The Note is in both. I've now added a second paragraph for this situation, and mentioned the note when we later talk about most of the options being negotiated.

r19643.