Hi Folks, unfortunately i noticed that the following NFS problem (i reported this already for RH 6.0) *still* exists. My Linux boxes (auto)mount NFS shares from a SUN/Solaris 2.5 Server (Solaris uses NFS v3). Serious errors occure when working on this NFS mounted shares (e.g. home directory), since Linux seems not to be able to correctly access NFS v3 shares (not even if a specify rsize=8192, nfs=v3, ...) I know this bug was reported several times before and someone told me knfsd gets fixed, but *when* will this bug be solved. RedHat 6.0 and 6.1 is *not* usable in a heterogenous environment. I had to re-install RH 5.2 (where this bug does not occure). Many Thanks Thomas (Thomas.Mandl.at)
It seams, that this problem *does not* appear when using a Solaris 7 (SunOS 5.7) NFS Server (also NFS v3).
We are running RH6.0 (stock) and mounting disk from a sun netra (SunOS nfs1 5.5.1 Netra_NFS_SmartServe_1.1_fcs sun4u sparc SUNW,Ultra-1). on process4 (RH6.0) I mount a (/u) directory from nfs1 (the netra). If I do a (in /u/clients/test): foreach loop (T*.zip) zoo -add test.zip $loop end zoo -test test.zoo The archive ends up being corrupt (I preverified all of the zip files). Each zip file (23 of them) is approx 4M in size.
This problem affect NFS servers which are AIX based also. Isnt it about time that Linux NFS became credible/useable if it is to be taken seriously?
We use solaris x86 nfs server (2.5,2.6,5.7) and have not seen this problem, yet.
patches for Solaris have been available for sun for over a year now. This is a Sun problem
The same bug found in Solaris also exists in AIX (and I would guess other OS'es as well). We found it in AIX 4.3.2. IBM has a fix for it which is a port of the Sun Fix. Sunsolve id is 4071076, IBM APAR Number is IY06322. In our case, it was object files getting corrupted during compiles. After patching our AIX box, the problem went away.