Description of problem: "cp a /mnt/b" produces "cannot create regular file ... Input/output error" when copying from RHEL 3.9 (2.4.21-52.ELsmp) to RHEL 5.1 and /mnt/b is NFS V3 mounted. /mnt/b is created as a zero byte file (i.e. file is created but contents are not copied and the I/O error is produced). Error log on RH 3 system shows: kernel: call_verify: server accept status: 2 last message repeated 2 times kernel: RPC: garbage, exit EIO after each failure. If a second attempt to do the "cp" command is made after the zero byte file is created, it works correctly, presumably because the file has already been created. Version-Release number of selected component (if applicable): RHEL 3.9, 2.4.21-52.ELsmp How reproducible: cp filea /mnt/fileb where, /mnt is an NFS v3 file system mounted on a remote RH 5.1 (2.6.18-53.1.6.el5) system and /mnt/fileb does not exist prior to the "cp" command. "cp" command is performed on a RH 3.9 (2.4.21-52.ELsmp) system. Steps to Reproduce: 1. See "How reproducible" above. 2. 3. Actual results: cp: cannot create regular file `/mnt/fileb': Input/output error In /var/log/messages: kernel: call_verify: server accept status: 2 last message repeated 2 times kernel: RPC: garbage, exit EIO A zero byte file is created for file "/mnt/fileb". Expected results: No errors and the file is copied. Additional info: Doesn't seem to be a problem with RH 4 or RH 5 systems. I also tried earlier versions of RH 3 (2.4.21-32.ELsmp and 2.4.21-27.0.1.ELsmp) and they don't have the problem either.
Hi Aurthur - Thanks for the feedback. It looks like this is a problem on the rhel5 nfs server side. *** This bug has been marked as a duplicate of 429109 ***