🚀 go-pugleaf

Hi!
We've had this problem for some time now, and it hits users from time
to time, so I thought I'd ask whether anybody has got any patch for this.
Scenario:
A Solaris machine mounts a filesystem from an Ultrix box. Suddenly things
just stop (in one directory), eventually the user will get those 'NFS
server xxx not responding'. truss says it hangs in a getdents() call 
(even says 'sleeping'). The same thing can be seen on the network, 
the solaris machine sends continously readdir requests, but the Ultrix
machine doesn't respond:
14:57:57.290415 solaris.35af4d8d > ultrix.nfs: 124 readdir [|nfs] (DF)

The following might be the same problem, but it seems very strange:
du: ./.netscape-cache/cache30EBFC8D016: No such file or directory
du: ./.netscape-cache/cache30F0F544003: No such file or directory
du: ./.netscape-cache/cache30F0F5E300B: No such file or directory
du: ./.netscape-cache/cache30F28DB2003: No such file or directory
du: ./.netscape-cache/cache30FDF873005: No such file or directory

the cache filenames have been truncated to 16 bytes, they are actually
24 characters long (I've also seen this for a 31 character filename).
Anyway, sometimes filenames just get truncated to 16 chars. 

The Solaris machine (2.4) has the following patches installed:
101753-01  101907-02  101945-10  101979-03  102007-01  102038-01  102216-02
101829-01  101920-01  101945-13  101981-01  102007-02  102044-01  102218-03
101878-01  101921-04  101945-32  101981-02  102011-02  102062-03  102292-01
101879-01  101922-04  101959-02  101983-01  102020-02  102070-01  102319-01
101880-03  101923-03  101959-06  102001-03  102035-01  102079-01  102680-03
101902-01  101925-01  101969-04  102002-01  102036-01  102112-01  102711-01
101905-01  101933-01  101975-01  102003-01  102037-01  102137-01  102922-03

The Ultrix machine patched versions of the objects sim_sched.o, sim_sm.o
and sim_94.o. (can't remember the code...)

The machines are in different subnetworks with a router in between. 
Originally I suspected the router, but I've seen this behaviour
on an Ultrix/Solaris pair on the same subnet, too.

Clues? Patches?

--harald
Harald Lundberg <hl@tekla.fi>;Tekla Oy,Koronakatu 1,FIN-02210,ESPOO,FINLAND
tel +358-{0-8879449work,0-8039489fax,0-8026752,11-2418013res,50-5578303mob)