On Thu, 26 Feb 2009, P. Larry Nelson wrote: > Hi all, while troubleshooting an odd NFS error, I discovered that apparently > (if you can believe the man pages) the default protocol for nfs clients to > mount from servers is now TCP. > > And it apparently started with SL4.7, tho I could find no mention of > such a default protocol change while perusing the release notes for > SL4.7. > > The following excerpts are from the man page for nfs(5) from > a 4.6 system and then from a 4.7 system. Note the change in the > default protocol. > > Under "Options for the nfs file system type" in the man page for nfs(5), > > ------------- > For SL4.6 (man page comes from util-linux-2.12a-17.el4_6.1): > > tcp Mount the NFS filesystem using the TCP protocol instead of the > default UDP protocol. Many NFS servers only support UDP. > ------------- > ------------- > For SL4.7 (man page comes from util-linux-2.12a-20.el4): > > tcp Mount the NFS filesystem using the TCP protocol. This is the default. > ------------- > > I am currently going thru and adding "udp" to all the SL4.7 clients' fstab > entries so they will use UDP rather than TCP. > > My main question is, lacking any explicit protocol designation in the fstab, > how can one tell which protocol a client is using? > mount > And lastly, why wasn't the change documented in the release notes? In SL.documention/RELEASE-NOTES-x86-en it states that TCP is now the default. Note that the RELEASE-NOTES-x86-en were the original release notes for SL 4.0 from TUV. We do not write the TUV release notes so I do not know why it was not documented. > > From what I've gleaned about the two protocols from googling, it appears > that TCP has advantages on a lossy network but that's not our scenario. > It also is not a stateless protocol, like UDP, so if a server crashes in > the middle of a packet transmission, the client will hang and filesystems > will need to be unmounted and remounted. So it would seem UDP is better, > at least in our case. > > Thanks! > - Larry > -connie sieh