Amanda-Users

Re: all estimate failed error on amanda client

2005-01-09 01:11:47
Subject: Re: all estimate failed error on amanda client
From: Gene Heskett <gene.heskett AT verizon DOT net>
To: David Newman <dnewman AT networktest DOT com>
Date: Sun, 9 Jan 2005 00:58:28 -0500
On Sunday 09 January 2005 00:09, David Newman wrote:
>On 1/8/05 3:14 PM, "Gene Heskett" <gene.heskett AT verizon DOT net> wrote:
>> On Saturday 08 January 2005 17:19, David Newman wrote:
>>> On Fri, 7 Jan 2005, Gene Heskett wrote:
>>>> On Thursday 06 January 2005 21:25, David Newman wrote:
> >>>> Greetings. I have installed Amanda on two FreeBSD 5.2.1 boxes,
>>>>>
>>>>> Host A and Host B. Host A is the Amanda server, and has the
>>>>> amanda-server package installed from ports. Host B has the
>>>>> amanda-client package. I've added a couple of Host B's
>>>>> partitions in Host A's disklist.
>>>>>
>>>>> I'm getting this error in the nightly Amanda reports:
>>>>>
>>>>> hostb /usr/home lev 0 FAILED [disk /usr/home, all estimate
>>>>> failed] hostb /var lev 0 FAILED [disk /var, all estimate
>>>>> failed]
>>>>>
>>>>> The backups on host A are working fine. On the client (host B)
>>>>> I've followed the directions here:
>>>>>
>>>>> http://amanda.sourceforge.net/docs/install.html#id2539781
>>>>>
>>>>> There is no firewall or other external impediment between the
>>>>> two boxes.
>>>>>
>>>>> Both boxes run amanda as user "operator" which by default in
>>>>> FreeBSD has no home directory. I have placed identical
>>>>> .amandahosts files in the root directory of both boxes.
>>>>>
>>>>> Any clues as to why backups of Host B are failing?
>>>>>
>>>>> thanks
>>>>>
>>>>> dn
>>>>
>>>> Are there any logs being generated on host B from amanda's
>>>> activities?
>>>>
>>>> You may have to change the operator to a user with a valid home
>>>> dir. I've always, from day one, had a user amanda, who was a
>>>> member of group disk.
>>>
>>> Thanks, I think this was the problem.
>
>Sorry, spoke too soon. The new configuration with amanda users
> passed amcheck but attempts to dump the client machine ("Host B"
> from my first email) still result in the "all estimate failed"
> error.
>
>In disklist on the amanda server, I specify directories rather than
>partition names to be backed up:
>
>hostb.example.com /dir1 comp-user
>hostb.example.com /dir2 comp-user
>
>And on the client machine, the debug logs in /tmp/amanda are
> complaining that /dir1 is unknown:
>
>sendsize[5334]: time 0.015: calculating for device '/dir1' with
>''
>sendsize[5334]: time 0.015: running "/sbin/dump 0Shsf 0 1048576 -
> /dir1" sendsize[5334]: time 0.017: running
> /usr/local/libexec/amanda/killpgrp sendsize[5334]: time 0.044:
> dump: /dir1: unknown file system
>
>Does amanda only back up entire partitions and not individual
> directories?

dump can only backup entire filesystems.  If /dir1 is just a 
directory, then you must use gnutar.

>Curtis Preston's backup book suggests (p. 164) that disklist can
> contain logical names like /usr (which in many Linux distros would
> be a directory and not a separate partition).  But the amanda
> manpage says it's gotta be a mount point.
>
>So -- does amanda work only on whole partitions, or is there some
> way I can use it to back up only selected directories?

As above, dump works only on whole filesystems from the mountpoint, 
tar can reach into subdirs and do most anything you want.  The tar 
version should be either 1.13-19 or 1.13-25, its known that the new 
1.14 is broken.
>
>many thanks
>
>dn
>
>
>
>
>from amandad.20050108140343000.debug:
>
>Amanda 2.4 REP HANDLE 001-005D0508 SEQ 1105221838
>OPTIONS features=fffffeff9ffe0f;
>/dir1 0 SIZE -1
>/dir2 0 SIZE -1
>----
>
>from sendsize.20050108140343.debug:
>
>sendsize: debug 1 pid 5332 ruid 1021 euid 1021: start at Sat Jan  8
> 14:03:43 200
>5
>sendsize: version 2.4.4p4
>sendsize[5332]: time 0.009: waiting for any estimate child: 1
> running sendsize[5334]: time 0.010: calculating for amname '/dir1',
> dirname '/dir1', spindle -1
>sendsize[5334]: time 0.010: getting size via dump for /dir1 level 0
>sendsize[5334]: time 0.015: calculating for device '/dir1' with
>''
>sendsize[5334]: time 0.015: running "/sbin/dump 0Shsf 0 1048576 -
> /dir1" sendsize[5334]: time 0.017: running
> /usr/local/libexec/amanda/killpgrp sendsize[5334]: time 0.044:
> dump: /dir1: unknown file system sendsize[5334]: time 0.046: .....
>sendsize[5334]: estimate time for /dir1 level 0: 0.030
>sendsize[5334]: no size line match in /sbin/dump output for "/dir1"
>sendsize[5334]: .....
>sendsize: debug 1 pid 5332 ruid 1021 euid 1021: start at Sat Jan  8
> 14:03:43 200
>5
>sendsize: version 2.4.4p4
>sendsize[5332]: time 0.009: waiting for any estimate child: 1
> running sendsize[5334]: time 0.010: calculating for amname '/dir1',
> dirname '/dir1', spindle -1
>sendsize[5334]: time 0.010: getting size via dump for /dir1 level 0
>sendsize[5334]: time 0.015: calculating for device '/dir1' with ''
>sendsize[5334]: time 0.015: running "/sbin/dump 0Shsf 0 1048576 -
> /dir1" sendsize[5334]: time 0.017: running
> /usr/local/libexec/amanda/killpgrp sendsize[5334]: time 0.044:
> dump: /var/named/etc/namedb: unknown file system sendsize[5334]:
> time 0.046: .....
>sendsize[5334]: estimate time for /dir1 level 0: 0.030
>sendsize[5334]: no size line match in /sbin/dump output for "/dir1"
>sendsize[5334]: .....
>sendsize[5334]: estimate size for /var/named/etc/namedb level 0: -1
> KB sendsize[5334]: time 0.046: asking killpgrp to terminate
>sendsize[5334]: time 1.051: done with amname '/dir1', dirname
> '/dir1', spindle -1
>sendsize[5332]: time 1.052: child 5334 terminated normally
>
>...and then the same thing again for /dir2.
>
>What's going on? The tape operator has read permissions on both
> /dir1 and /dir2. And

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.31% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attorneys please note, additions to this message
by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.