I can see a stale entry in the output of dfmounts. My troublesome client seems to show up as a client of the filesystem, but it doesn't really have that file system mounted on the client side. My theory is this stale connection is interfering with a new attempt to mount that filesystem from the client.
Is there any way (on the server) to reset/flush this record without rebooting the system? --chris Chris Quenelle wrote: > I'm having some funky NFS mount behavior using the latest internal > opensolaris build (b101). My b92 machine works fine on the same > mount points. > > A few random filesystems are not mountable or get mounted readonly > when they shouldn't. > > Does anyone have any simple dtrace scripts, or recipes for turning > on diagnostic output? Focussing on the authentication step. > > It fails using V3 and V4. > > One mount point works and shows this: > tiaga.SFBay.Sun.COM -> ilm-nfs.SFBay.Sun.COM NFS C 4 (secinfo ) PUTROOTFH > SECINFO compilers > ilm-nfs.SFBay.Sun.COM -> tiaga.SFBay.Sun.COM NFS R 4 (secinfo ) NFS4_OK > PUTROOTFH NFS4_OK SECINFO NFS4_OK AUTH_SYS > tiaga.SFBay.Sun.COM -> ilm-nfs.SFBay.Sun.COM NFS C 4 (mount ) PUTROOTFH > GETFH LOOKUP compilers GETFH GETATTR c8000167 0 > ilm-nfs.SFBay.Sun.COM -> tiaga.SFBay.Sun.COM NFS R 4 (mount ) NFS4_OK > PUTROOTFH NFS4_OK GETFH NFS4_OK FH=4CD1 LOOKUP NFS4_OK GETFH NFS4_OK FH=8C97 > GETATTR NFS4_OK > > Another mount point fails with this: > tiaga.SFBay.Sun.COM -> ilm-nfs.SFBay.Sun.COM NFS C 4 (secinfo ) PUTROOTFH > LOOKUP compilers SECINFO train_src > ilm-nfs.SFBay.Sun.COM -> tiaga.SFBay.Sun.COM NFS R 4 (secinfo ) NFS4_OK > PUTROOTFH NFS4_OK LOOKUP NFS4_OK SECINFO NFS4_OK > tiaga.SFBay.Sun.COM -> ilm-nfs.SFBay.Sun.COM NFS C 4 (mount ) PUTROOTFH > GETFH LOOKUP compilers GETFH GETATTR c8000167 0 LOOKUP train_src GETFH > GETATTR c8000167 0 > ilm-nfs.SFBay.Sun.COM -> tiaga.SFBay.Sun.COM NFS R 4 (mount ) > NFS4ERR_ACCESS PUTROOTFH NFS4_OK GETFH NFS4_OK FH=4CD1 LOOKUP NFS4_OK GETFH > NFS4_OK FH=8C97 GETATTR NFS4_OK LOOK... > > > But that doesn't tell me much except the mount failed. > > --chris