I received one response to my rdump question (included below).
Thanks Phil! His suggestion is a good work around but I still
do not have an answer to my original post (also included below).
Bob
Sisk
Phil wrote:
>If you know that your dump will fit on the remote tape, then I suggest
>discarding "rdump" altogether and using a "dump | rsh dd" pipe.
>For example, if the tape drive is on machine "sun" and you are
>dumping from machine "du", log into "du" as root and run this command.
>This requires that you have the correct .rhosts file for root on both
>machines (which I think you need for rdump, too, anyway).
>
> dump 0uf - | rsh sun dd obs=10k of=/dev/tape_device
>
>10k is the standard block size for dump tapes. You could try
>increasing that, say to 60k, for better tape performance. But
>then you have to remember to provide a block size option when
>you try to restore from the tape.
>
>Years ago, I did some timed tests of rdump versus the dump | rsh dd
>pipeline on 4.3BSD UNIX systems. The dump | rsh pipeline was nearly
>three times as fast! Of course, with modern faster computers
>and perhaps improved software that may no longer be true. In any
>case, I am sure that the dump | rsh pipeline is not any slower
>than rdump.
>
>-Phil Farrell, Computer Systems Manager
>Stanford University School of Earth Sciences
>farrell_at_pangea.stanford.edu
Original Question:
>>Ok here is the problem. I have a solaris 2.6 machine with
>>an "Advanced Intelligent Tape" drive attached for backups.
>>The AIT tape drive is made by Sony and has a Native
>>capacity of 25 GB/per tape. I want to do a remote dump of
>>the drives from my du 4.0c machine to the AIT drive on the sun.
>>From man rdump on the du machine:
>>
>> Although the rdump command can detect magnetic tape on remote ULTRIX and
>> Digital UNIX systems, it cannot detect magnetic tape on other remote sys-
>> tems. Instead, it defaults to treating the output medium as a disk file
>> and aborts the operation should it encounter overflow or I/O error cases.
>>
>>When I run rdump (du -> sun) I get the following:
>>
>># rdump -0uf sun_1:/dev/rmt/0n /
>>rdump: Dumping from host du_1
>>rdump: Date of this level 0 dump: Mon Sep 21 16:29:17 1998 EDT
>>rdump: Date of last level 0 dump: the start of the epoch
>>rdump: Dumping /dev/rrz16a (/) to /dev/rmt/0n on host sun_1
>>rdump: Mapping (Pass I) [regular files]
>>rdump: Mapping (Pass II) [directories]
>>rdump: Remote system is not known to be compatible for magtape functions, output being treated as regular file
>>rdump: Estimate: 107718 tape blocks on disk
>>The available blocks(6918),is less than the estimated blocks(107718)
>>Dump is being aborted
>>rdump: SIGTERM received
>>rdump: The ENTIRE dump is aborted
>>
>>Is there a way to get around this "not known to be compatible for magtape
>>functions" mess. I have tried to specify what I think are the correct
>>block per write (-b), density (-d), and tape size (-s) values on the command
>>line. When I do this I always get the following error:
>>
>>rdump: Cannot set socket send-buffer size
>>rmthost(): setsockopt(): Error 0
>>Terminated
>>
>>Any help would be appreciated. I will post a summary.
>>
>>Bob
>>Sisk
>>rsisk_at_marine.unc.edu
Received on Wed Sep 30 1998 - 12:41:48 NZST