A few days ago I had asked about NFS write errors 69 and 70.
Thanks to the following who took the time to reply:
shollen_at_valhalla.cs.wright.edu
Peter Stern <peter_at_wiscpa.weizmann.ac.il>
mogul_at_actitis.pa.dec.com (Jeffrey Mogul)
Bob Jones <BJ_at_OREGON.UOREGON.EDU>
My original question and the replies follow:
Hello,
We are running DU version 3.2 and 3.2C utilizing NFS.
On some of the clients we get the following errors:
Apr 17 18:09:01 my_client vmunix: NFS3 write error 69 on host
my_server.mydomain
Apr 17 18:09:01 my_client vmunix: UBC write error(69) on device (0, 0) at
block 0 size 8192
On another client we get :
Apr 17 08:54:23 my_client2 vmunix: NFS3 write error 70 on host
my_server.mydomain
Apr 17 08:54:23 my_client2 vmunix: UBC write error(70) on device (0, 0) at
block 0 size 8192
Yet on another client I get:
Apr 17 17:02:08 my_client3 vmunix: NFS3 write error 69 on host
my_server.mydomain
On this client there is no UBC write errors.
Still on other clients there are no errors. Also, no errors are
reported on the server.
My question is what do error numbers mean? how can I get rid of them?
I feel NFS performance has been degraded, but I can't pin point an
exact problem. Could it be my hard disk on the server is going bad?
I don't see anything in uerf or other logs.
Thanks in advance
Kevin Behrooz
National Superconducting Cyclotron Lab
East Lansing, Michigan
---------------------------
Complete replies I have received:
>From shollen_at_valhalla.cs.wright.edu Fri Apr 18 07:31:34 1997
These are nfs errors, and can be found in /usr/include/nfs/nfs.h:
valhalla nfs> grep 69 *.h
nfs.h: NFS3ERR_DQUOT = 69,
valhalla nfs> grep 70 nfs.h
NFS3ERR_STALE = 70,
Error 69 indicates that someone on the client is trying to exceed their quota,
70 is a stale file handle, which could mean that someone is trying to write to
a file which is no longer there.
From: Peter Stern <peter_at_wiscpa.weizmann.ac.il>
You can find the error nos, in:
/usr/include/errno.h
#define EDQUOT 69 /* Disc quota exceeded */
#define ESTALE 70 /* Stale NFS file handle */
The former is obvious. The latter may be solved by re-nount-ing the
disk. Could also be Netscape trying to point to cache which has
been removed.
From: mogul_at_actitis.pa.dec.com (Jeffrey Mogul)
The error numbers correspond to those listed in /usr/include/errno.h
error 69 is "Disc quota exceeded" ... the cause of this should
be clear.
error 70 is an "NFS error code" named ESTALE, which presumably
corresponds to "stale file handle".
-Jeff
From: Bob Jones <BJ_at_OREGON.UOREGON.EDU>
> My question is what do error numbers mean? how can I get rid of them?
Most if not all of the NFS errors are listed (very briefly) in
/usr/include/nfs/nfs_prop.h. I use the following alias to check NFS errors:
alias nfs_err 'grep "ERR_.* = \!:1" /usr/include/nfs/nfs_prop.
Incidentally, if you also have any Solaris systems, the errors are listed in
/usr/include/nfs/nfs.h under Solaris 2.x.
Received on Mon Apr 21 1997 - 14:55:21 NZST