I'm trying to work out why one of my machines can't run NFS in a lipkg zone. The nfs/server svc fails because it can't run mountd, and if I run that manually, it exits 1. Comparing a mountd truss from a working machine, everything looks the same up until a name_service_door call, after which the working one looks at nss.conf, and the broken one doesn't. (As shown in the gist at gist.github.com/snltd/28639add2108d0a21b4646e7a538f089) I

can't find anything different in the config of the working and non-working hosts, in either the global or local zones. Does anyone have any suggestions? They're both on the same revision of omnios-r151044.

is nss setup to use DNS? default from illumos is files only if omnios hasn't changed that.

Yes. Everything else works as expected in every other zone on the box, but no zone can start mountd.

blepe: I would make sure, that the own local name and IP are correctly resolved back and forward, check with the tool "getent"

getent hosts thezonemachinename

and getent hosts ip_of_zonemachinename

That was it! The DNS for the new box hadn't propagated. I didn't realise NFS and DNS were quite so tightly tangled. Thank you.

NFS and DNS are tangled if you use DNS :) Alternatives are /etc/hosts or other nss providers that are configured in /etc/nsswitch.conf

I'm getting NIS+ flashbacks!

no LDAP flashbacks?

No, that's always been someone else's problem :)

XD

blepe: glad to have helped. stumbled more then once into this little oddity with SunOS/Illumos based NFS serving

thanks for givinge

me ldapclient and dns flashbacks