Re: your mail

[Matt Hyclak]

On Tue, Jul 19, 2005 at 10:48:23AM -0600, [EMAIL PROTECTED] enlightened us:
> Well.
> I have checked the docs and faq.
> 
> 
> I'm running the command as root as specified by the faq
> http://amanda.sourceforge.net/cgi-bin/fom?_recurse=1&file=5
> 
> 
> >You see, I had 127.0.0.1 configured for user amanda for amdump. Amrecover HAS
> to be run as root.
> 
> 
> Also running the command specifying index server and tape server fails in the
> same way as if it were localhost.
> 
> #amrecover CCC -s fastkernel.chem.utah.edu -t fastkernel.chem.utah.edu
> AMRECOVER Version 2.4.4p1. Contacting server on fastkernel.chem.utah.edu ...
> amrecover: cannot connect to fastkernel.chem.utah.edu: Connection refused
> [EMAIL PROTECTED] xinetd.d]# cat /var/lib/amanda/.amandahosts
> ##127.0.0.1  root
> fastkernel.chem.utah.edu root
> 
> Thanks for your help
> Rich

It looks like amindexd isn't listening. What is your xinetd configuration
for amanda?

Matt

-- 
Matt Hyclak
Department of Mathematics 
Department of Social Work
Ohio University
(740) 593-1263

Re: your mail

[Jon LaBadie]

On Sun, May 22, 2005 at 09:53:14AM -0300, vlpg wrote:
> Hi all,
> 
> I would like to know what port the client is listening for the server amanda.
> Because I have open the firewall for the client at the port 10080, but it 
> isn't working.
> 

This is described in a document "PORTS.USAGE"
in the amanda docs directory.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

Re: your mail

[Matt Hyclak]

On Sun, May 22, 2005 at 09:53:14AM -0300, vlpg enlightened us:
> I would like to know what port the client is listening for the server amanda.
> Because I have open the firewall for the client at the port 10080, but it 
> isn't working.
> 
> 

Initially it does listen on 10080, but then opens up other connections. Read
PORT.USAGE (or portusage.txt in 2.4.5) for more details. If you are using an
iptables firewall, you can likely get by with inserting the
ip_conntrack_amanda module. See my post from the last week or so regarding
that.

Matt

-- 
Matt Hyclak
Department of Mathematics 
Department of Social Work
Ohio University
(740) 593-1263


pgp6nDebSWhR6.pgp
Description: PGP signature

Re: your mail

[Jon LaBadie]

On Wed, Apr 27, 2005 at 06:38:51PM -0400, McDonagh, Joe wrote:
> Hi, I am trying to get Amanda going on a RedHat ES 3 box, but I am
> having this same issue addressed on a message board when trying to back
> up a directory that is remote via NFS:
>  
> "...I was then able to do an amcheck, and it succeeded. Doing an amdump
> still
> fails however. Looking into the log file (sendsize.-.debug) it
> indicated that
> amdump is not receiving the size information (presumably from sendsize
> module)
> that it needs to complete the tar"
>  
> 
> http://www.mail-archive.com/amanda-users@amanda.org/msg23489.html
>  
> The responses say it's a problem with GNU tar 1.13, and to use 1.13-25,
> however that is the version that the redhat box is running. What am I
> missing?
> 

The reason why you have not gotten size information.  :)

There are many reasons why you might not get a valid response.
The version of gtar might have been involved in the message board
posters problem and have no relation to your problem.

There are probably other indications what was happening if you
did deeper into those log files/reports.  Both on client and
server.

BTW one common problem for large DLE's is getting the various
timeouts setup properly.  Maybe the estimate was just taking
too darn long and the server gave up waiting.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

Re: your mail

[Joshua Baker-LePain]

On Mon, 4 Apr 2005 at 3:15pm, Andrés Fernando Argüelles Delgado wrote

> |   DUMP: You can't update the dumpdates file when dumping a subdirectory
> |   DUMP: The ENTIRE dump is aborted.
> sendbackup: error [/sbin/dump returned 1]
> \

>  I do not know why amanda returns that

RTFEM.  You're trying to use dump to backup something that isn't a 
full filesystem.  Don't do that.  Use tar instead.

And search the archives, this is a FAQ.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Jon LaBadie]

On Mon, Apr 04, 2005 at 01:44:20AM -0500, Andrés Fernando Argüelles Delgado 
wrote:
> I finally made my amanda configuration work, The amcheck results are
> fine, but when I run amdump for a test it sends to
> /var/lib/amanda/dead.letter a error report:
> 
> >From amanda Mon Apr  4 01:36:18 2005
> Return-Path: 
> Received: (from [EMAIL PROTECTED])
>   by localhost (8.13.1/8.13.1/Submit) id j346aII0010390
>   for amanda; Mon, 4 Apr 2005 01:36:18 -0500
> Date: Mon, 4 Apr 2005 01:36:18 -0500
> From: Amanda user 
> Message-Id: <[EMAIL PROTECTED]>
> To: amanda
> Subject: DailySet1 AMANDA MAIL REPORT FOR April 4, 2005
> 
> These dumps were to tape Etiqueta.
> The next tape Amanda expects to use is: Etiqueta.
> 
> FAILURE AND STRANGE DUMP SUMMARY:
>   driver: FATAL exec /usr/local/libexec/dumper (dumper0): Permission denied
>   driver: FATAL exec /usr/local/libexec/dumper (dumper1): Permission denied
>   driver: FATAL exec /usr/local/libexec/dumper (dumper2): Permission denied
>   driver: FATAL exec /usr/local/libexec/dumper (dumper3): Permission denied

When you installed amanda, i.e. did "make install",
were you running as root?  You should have been.

I'm guessing that 'dumper', which should be executable, owned by root,
and set-uid, does not have the correct permissions.  Doing the install
as root should set them.


-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

Re: your mail

[Matt Hyclak]

On Thu, Mar 31, 2005 at 06:43:58PM -0500, Andrés Fernando Argüelles Delgado 
enlightened us:
> I followeb the configuration for making backups on disk instead on
> tape described in the Amanda-Tape-dirves file but I get the next error
> when I run amlabel.
> 
> Couldn´t load tapelist from /usr/local/etc/amanda/DailySet1 no such
> file or directory
> 
> I know, according amanda man page, that this file is generated for
> amanda automatically when you run amlabel. How do I fix this problem?
> 

touch /usr/local/etc/amanda/DailySet1/tapelist and chown amanda (or whatever
user amanda runs as)

-- 
Matt Hyclak
Department of Mathematics 
Department of Social Work
Ohio University
(740) 593-1263

Re: your mail

[Jon LaBadie]

On Thu, Mar 31, 2005 at 06:43:58PM -0500, Andrs Fernando Argelles Delgado wrote:
> I followeb the configuration for making backups on disk instead on
> tape described in the Amanda-Tape-dirves file but I get the next error
> when I run amlabel.
> 
> Couldn´t load tapelist from /usr/local/etc/amanda/DailySet1 no such
> file or directory
> 
> I know, according amanda man page, that this file is generated for
> amanda automatically when you run amlabel. How do I fix this problem?
> 
> 
Try creating an empty tapelist file and rerun amlabel.

I think it must exist first.
Either as empty or with a single blank line.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

Re: your mail

[Jon LaBadie]

On Tue, May 11, 2004 at 08:50:53AM -0400, Jonathan Dill wrote:
> Gavin Henry wrote:
> 
> I have periodically searched for ufsrestore-compatible software for 
> Linux, but have been unable to find any yet.  In my case, I wanted to be 
> able to index dumps from a Solaris amanda client on a Linux amanda server.
> 

As the Sun's file system UFS is, or is based on, the BSD filesystem,
might one of the BSD tools be compatible?  Don't know, just asking.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

RE: your mail

[Gavin Henry]

>A few ideas, which might be useful for the future at least:  1) switch 
>to GNUTAR (I was using GNUTAR but had ended up switching to ufsdump due 
>to some problems); 2) there used to be a "binary compatibility layer" 
>that let you execute Solaris executables on Linux, but maybe that has 
>disappeared, or at least would probably require compiling a "custom 
>kernel"; 3) look for source code for ufsrestore, but I don't think it's 
>Open Source; 4) Can Solaris run under VMware on x86 architecture? 5) 
>look for Open Source libraries that can parse ufsdump format, or ufs 
>filesystem, and write your own code.

Thanks for this. I should of clicked that the command listed in the docs 
wasn't for tar. I do use gnutar, so the command would be:

dd if=/dev/nst0 bs=32k skip=1 | tar xzvf -

Re: your mail

[Jonathan Dill]

Gavin Henry wrote:

Sorry, forgot to put in a subject.

Yes, Fedora. Is there anything else like it for Linux?
 

I have periodically searched for ufsrestore-compatible software for 
Linux, but have been unable to find any yet.  In my case, I wanted to be 
able to index dumps from a Solaris amanda client on a Linux amanda server.

A few ideas, which might be useful for the future at least:  1) switch 
to GNUTAR (I was using GNUTAR but had ended up switching to ufsdump due 
to some problems); 2) there used to be a "binary compatibility layer" 
that let you execute Solaris executables on Linux, but maybe that has 
disappeared, or at least would probably require compiling a "custom 
kernel"; 3) look for source code for ufsrestore, but I don't think it's 
Open Source; 4) Can Solaris run under VMware on x86 architecture? 5) 
look for Open Source libraries that can parse ufsdump format, or ufs 
filesystem, and write your own code.

If #2 still exists (though I don't think so) there might be a Linux 
distro out there tailored to exploit that capability.

--jonathan

RE: your mail

[Gavin Henry]

>If you need cross-platform capabilities, then 
>use tar for your backups rather than dump.

oh, I do use gnutar for my backups.

I should just replace ufsrestores with tar then. Silly me.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

RE: your mail

[Joshua Baker-LePain]

Please put responses below quoted text -- it makes the conversation much 
easier to follow.  Fixed below.

On Tue, 11 May 2004 at 11:36am, Gavin Henry wrote
> From: Joshua Baker-LePain [mailto:[EMAIL PROTECTED]

> > Err, by RPM I'm assuming you mean for a Linux system, and by 
> > ufsrestore/ufsdump I'm assuming you mean the Solaris backup utilties, 
> > which I'm pretty sure haven't been ported to Linux.  You can still search 
> > through a tape by using 'dd' to read the file headers.  When you find the 
> > right image, 'dd' it off the tape, send it to a Solaris box, and use 
> > ufsrestore on it.
> 
> Yes, Fedora. Is there anything else like it for Linux?

You could try just plain 'restore', but I highly doubt it'll be able to 
read Sun's ufsdump format.  If you need cross-platform capabilities, then 
use tar for your backups rather than dump.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

RE: your mail

[Gavin Henry]

Sorry, forgot to put in a subject.

Yes, Fedora. Is there anything else like it for Linux?

-Original Message-
From: Joshua Baker-LePain [mailto:[EMAIL PROTECTED]
Sent: 11 May 2004 11:29 AM
To: Gavin Henry
Cc: [EMAIL PROTECTED]
Subject: Re: your mail


On Tue, 11 May 2004 at 8:44am, Gavin Henry wrote

> Anyone know where I can get an RPM with ufsrestore/ufsdump in it, so I 
> can search through a tape as per the RESTORE file in the Amanda docs?

Err, by RPM I'm assuming you mean for a Linux system, and by 
ufsrestore/ufsdump I'm assuming you mean the Solaris backup utilties, 
which I'm pretty sure haven't been ported to Linux.  You can still search 
through a tape by using 'dd' to read the file headers.  When you find the 
right image, 'dd' it off the tape, send it to a Solaris box, and use 
ufsrestore on it.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Joshua Baker-LePain]

On Tue, 11 May 2004 at 8:44am, Gavin Henry wrote

> Anyone know where I can get an RPM with ufsrestore/ufsdump in it, so I 
> can search through a tape as per the RESTORE file in the Amanda docs?

Err, by RPM I'm assuming you mean for a Linux system, and by 
ufsrestore/ufsdump I'm assuming you mean the Solaris backup utilties, 
which I'm pretty sure haven't been ported to Linux.  You can still search 
through a tape by using 'dd' to read the file headers.  When you find the 
right image, 'dd' it off the tape, send it to a Solaris box, and use 
ufsrestore on it.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Joshua Baker-LePain]

On Sat, 21 Feb 2004 at 1:31pm, Arif Ali wrote

> 
> I am sorry about that, I was in a quite of a rush last night. 
> 
> amrecover for some reason is looking for hostname "linux" instead of 
> "backup-server". Is there an underlying program that amrecover uses from 
> which it gets it's configuration regarding the hostnames from.

It gets it at compile time.  You can override the compile time default on 
the command line.  See 'man amrecover'.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Jon LaBadie]

On Tue, Aug 19, 2003 at 09:28:54AM +0200, Wimmers, Heinrich wrote:
> Hi,
> 
> why do my amflush needs over 99% of my CPU-resource? I am using RedHat 9.0,
> and a HP5683A DAT-Drive. Oh, BTW, there are 2 SCSI Controllers in that
> machine: 1Adaptec 2100S with 5 HDD (RAID 5 and 1 HOTSPARE), and 1 Adaptec
> 19160 for the DAT-Drive. 
> 

It certainly doesn't here.  Solaris (dual cpu),
single SCSI controller (dual channel) and a DAT drive.

Amflush is a very low impact activity.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

Re: your mail

[Paul Bijnens]

Paul Bijnens wrote:
An example of a wrapper on the client...
Set the exitcodes correct too!
Here is an untested example.

#!/usr/bin/perl

use Getopt::Long qw(:config pass_through);

$result = GetOptions (
'directory=s' => \$dir,
'file=s' => \$file,
);
$post = 0;
if ($dir eq '/my/special_dir' &&  $file ne '/dev/null') {
system '/usr/bin/dbshut thedb >/tmp/amanda/dbshut.debug 2>&1';
# avoid output to stdout!!! stdout fd = the backup image
# msgs on stderr will be flagged as "strange" in amreport
# so you could leave out the 2>&1
$post = 1;
}
# put all options we processed back
unshift(@ARGV, '--directory', $dir);
unshift(@ARGV, '--file', $file;
# now put the real gnutar program in front
unshift(@ARGV, "/usr/local/bin/gtar");
$rc = $? >> 8;

system @ARGV ;  # and run it

if ($post) {  # need postprocessing?
system '/usr/bin/dbstart thedb >/tmp/amanda/dbstart.debug 2>&1';
}
exit $rc;

--
Paul Bijnens, XplanationTel  +32 16 397.511
Technologielaan 21 bus 2, B-3001 Leuven, BELGIUMFax  +32 16 397.512
http://www.xplanation.com/  email:  [EMAIL PROTECTED]
***
* I think I've got the hang of it now:  exit, ^D, ^C, ^\, ^Z, ^Q, F6, *
* quit,  ZZ, :q, :q!,  M-Z, ^X^C,  logoff, logout, close, bye,  /bye, *
* stop, end, F3, ~., ^]c, +++ ATH, disconnect, halt,  abort,  hangup, *
* PF4, F20, ^X^X, :D::D, KJOB, F14-f-e, F8-e,  kill -1 $$,  shutdown, *
* kill -9 1,  Alt-F4,  Ctrl-Alt-Del,  AltGr-NumLock,  Stop-A,  ...*
* ...  "Are you sure?"  ...   YES   ...   Phew ...   I'm out  *
***

Re: your mail

[Paul Bijnens]

Brian Cuttler wrote:
I know you can use wrappers for doing similar things on the
amanda client (when the client != server) but I've never quite
gotten that down correctly and I'm looking to start/stop a
database on an amanda client so the files are quiessed during
the dump.
What is the right way to handle surounding tasks on the amanda-client ?
(A brief example if someone doesn't mind too much...)


Here is an untested example.

#!/usr/bin/perl

use Getopt::Long qw(:config pass_through);

$result = GetOptions (
'directory=s' => \$dir,
'file=s' => \$file,
);
$post = 0;
if ($dir eq '/my/special_dir' &&  $file ne '/dev/null') {
system '/usr/bin/dbshut thedb >/tmp/amanda/dbshut.debug 2>&1';
# avoid output to stdout!!! stdout fd = the backup image
# msgs on stderr will be flagged as "strange" in amreport
# so you could leave out the 2>&1
$post = 1;
}
# put all options we processed back
unshift(@ARGV, '--directory', $dir);
unshift(@ARGV, '--file', $file;
# now put the real gnutar program in front
unshift(@ARGV, "/usr/local/bin/gtar");
system @ARGV ;  # and run it

if ($post) {  # need postprocessing?
system '/usr/bin/dbstart thedb >/tmp/amanda/dbstart.debug 2>&1';
}
--
Paul Bijnens, XplanationTel  +32 16 397.511
Technologielaan 21 bus 2, B-3001 Leuven, BELGIUMFax  +32 16 397.512
http://www.xplanation.com/  email:  [EMAIL PROTECTED]
***
* I think I've got the hang of it now:  exit, ^D, ^C, ^\, ^Z, ^Q, F6, *
* quit,  ZZ, :q, :q!,  M-Z, ^X^C,  logoff, logout, close, bye,  /bye, *
* stop, end, F3, ~., ^]c, +++ ATH, disconnect, halt,  abort,  hangup, *
* PF4, F20, ^X^X, :D::D, KJOB, F14-f-e, F8-e,  kill -1 $$,  shutdown, *
* kill -9 1,  Alt-F4,  Ctrl-Alt-Del,  AltGr-NumLock,  Stop-A,  ...*
* ...  "Are you sure?"  ...   YES   ...   Phew ...   I'm out  *
***

Re: your mail

[Brian Cuttler]

Toomas,

Thanks, I'd seen similar for doing things like putting the tape
offline after the amdump completed.

I know you can use wrappers for doing similar things on the
amanda client (when the client != server) but I've never quite
gotten that down correctly and I'm looking to start/stop a
database on an amanda client so the files are quiessed during
the dump.

What is the right way to handle surounding tasks on the amanda-client ?
(A brief example if someone doesn't mind too much...)


> > Instead of just running amdump from crontab, run something to a tune of:
> > mount /backupdisk && amdump YourConfig && umount /backupdisk

thanks,

Brian

---
   Brian R Cuttler [EMAIL PROTECTED]
   Computer Systems Support(v) 518 486-1697
   Wadsworth Center(f) 518 473-6384
   NYS Department of HealthHelp Desk 518 473-0773

Re: your mail

[Jon LaBadie]

On Mon, Jun 02, 2003 at 10:29:15AM -0300, Roberto Samarone Araujo (RSA) wrote:
> 
>   [EMAIL PROTECTED] /sisges]# amrecover Mensal1 -s server1 -t server1
>   
>   Added /linux.apps/test1.sh
>   amrecover> extract
> 
>   Extracting files using tape drive /dev/null on host server1. 
   

Configuration problem.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

Re: your mail

[Joshua Baker-LePain]

On Mon, 24 Mar 2003 at 5:41pm, anne wrote

> # amrecover edge
> AMRECOVER Version 2.4.2p2. Contacting server on localhost ...
> 220 helium AMANDA index server (2.4.2p2) ready.
> 200 Access OK
> Setting restore date to today (2003-03-24)
> 200 Working date set to 2003-03-24.
> 200 Config set to edge.
> 501 No index records for host: helium. Invalid?
> Trying helium.edge-it.subnet ...
> 200 Dump host set to helium.edge-it.subnet.
> Can't determine disk and mount point from $CWD
> 
> what is this disk it ask for ?

Err, the disk you want to restore.  :)

Try, e.g.,  'setdisk /home' (or a corresponding disklist entry).

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Karl Hudnut]

Joshua:

You are correct, there is a docs directory in the 2.4.3 tarball and it has 
the PORTS.USAGE document. 

Thanks again. And sorry for disseminating missinformation.

--
Dr. Karl Hudnut   System Administrator  UCAR - COSMIC
[EMAIL PROTECTED]   http://www.cosmic.ucar.edu303 497 8024


On Mon, 24 Feb 2003, Joshua Baker-LePain wrote:

> CCed back to the list for the archives...
> 
> On Mon, 24 Feb 2003 at 3:01pm, Karl Hudnut wrote
> 
> > Thanks for your help. I don't seem to have that documentation in any of
> > the rpms I have here. It doesn's seem to be in the tarballs from
> > amanda.org either. Can you help me find it?
> 
> ??  I just downloaded amanda-2.4.3.tar.gz via the link 
>  
> on  and there's an entire directory 
> in there called 'docs'.
> 
> > I also wonder if you have any comment on amcheck succeeding and amdump 
> > failing.
> 
> Amcheck just connects to the amandad port (UDP 10080) on the client.  It 
> doesn't check that the client can actually open the TCP connections back 
> to the server.
> 
> Grepping through my own mail archives, here's a summary of port usage I 
> posted back in 2001:
> 
> The amanda server sends a UDP sendbackup request from a privileged port 
> (not necessarily the same one as above) to port 10080 on the client.
> The amanda client sends a UDP ACK from port 10080 to the originating 
> privileged port on the server.  It then sends another one containing the 
> numbers of three (non-privileged) TCP ports to set up the data, message, 
> and index connections.
> The amanda server sends a UDP ACK from the privileged port to port 10080 
> on the client.
> The amanda server then initiates three TCP connections on the ports 
> indicated in the UDP packet from the client.  These are on unprivileged 
> ports on both systems.  The dumper on the client then proceeds to start 
> sending date over the TCP connections.
> 
> -- 
> Joshua Baker-LePain
> Department of Biomedical Engineering
> Duke University
> 
> 

Re: your mail

[Joshua Baker-LePain]

CCed back to the list for the archives...

On Mon, 24 Feb 2003 at 3:01pm, Karl Hudnut wrote

> Thanks for your help. I don't seem to have that documentation in any of
> the rpms I have here. It doesn's seem to be in the tarballs from
> amanda.org either. Can you help me find it?

??  I just downloaded amanda-2.4.3.tar.gz via the link 
 
on  and there's an entire directory 
in there called 'docs'.

> I also wonder if you have any comment on amcheck succeeding and amdump 
> failing.

Amcheck just connects to the amandad port (UDP 10080) on the client.  It 
doesn't check that the client can actually open the TCP connections back 
to the server.

Grepping through my own mail archives, here's a summary of port usage I 
posted back in 2001:

The amanda server sends a UDP sendbackup request from a privileged port 
(not necessarily the same one as above) to port 10080 on the client.
The amanda client sends a UDP ACK from port 10080 to the originating 
privileged port on the server.  It then sends another one containing the 
numbers of three (non-privileged) TCP ports to set up the data, message, 
and index connections.
The amanda server sends a UDP ACK from the privileged port to port 10080 
on the client.
The amanda server then initiates three TCP connections on the ports 
indicated in the UDP packet from the client.  These are on unprivileged 
ports on both systems.  The dumper on the client then proceeds to start 
sending date over the TCP connections.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Joshua Baker-LePain]

On Mon, 24 Feb 2003 at 1:52pm, Karl Hudnut wrote

> Has anyone seen a situation in which amcheck succeeds yet amdump fails 
> with the following:
> 
> FAILURE AND STRANGE DUMP SUMMARY:
>   mail   /home lev 2 FAILED [could not connect to mail]
>   mail   /var/spool/mail lev 2 FAILED [could not connect to mail]
> 
> I know that this is an iptables problem. Because if I turn off iptables
> the dump succeeds. I have destination ports 10080 - 10083 open to the
> server for both udp and tcp packets in my iptables on the client:

You need more than that.  Look in docs/PORT.USAGE for details on how 
amanda allocates ports.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Jon LaBadie]

On Mon, Nov 11, 2002 at 08:02:37AM +0100, Nicolas Cartron wrote:
> 
> I want to install Amanda on a production environment,
> backup server on FreeBSD and clients running Linux, Solaris or FreeBSD.

I suspect you really mean "you are considering installing amanda".
> 
> I read in the section dedicated to Amanda (in the 'Unix backup & recovery',
> O'reilly) the following lines :
> 
> AMANDA currently starts a new tape for each run and does not provide a
> mechanism to append a new run to the same tape as a previous run
>  ... 
> Does that mean that if i want to do the same with Amanda, i have to
> rotate the tape every day ?

That's what it means.   Though you could collect multiple backups on
disk (aka holding disk) then "flush" all to tape at one time.

> We are currently running Brightstor from Computer Associates,
> doing a full backup every sunday, and incremental backups on ONE tape
> everyday.

Such a traditional way of backing up is exactly what amanda tries
to get away from.  Amanda mixes some full backups and some incrementals
each day with the aim of a consistant sized backup each day.

No more 18 hour, 300 GB dumps on Sunday and 2 hour 10 GB dumps other days.
Instead amanda does 60 GB dumps daily, including Sunday.


> (I forgot to explain that I have 2 DLT tapes backuping the datas).

Amanda can be folded, spindled, and mutilated in many ways.  It was not
designed to use multiple drives, but "I think" some are doing just that.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)

Re: your mail

[Jean-Francois Malouin]

* Eric Zylstra ([EMAIL PROTECTED]) [20020320 14:00] thus spake:
> I'm running into an old problem--this is in the archives from a few 
> years back.  I have done `rm config.*` and rerun configure with the same 
> result.
> 
> 
> Compiling amanda-2.4.2 (client only) on Irix 6.5.x, with gcc 3.0.3 and 
> gmake 3.78.1 (although behavior is the same under Irix's make):

hmm, I can compile amanda-2.4.2p2 with egcs-2.91.66 19990314 
(egcs-1.1.2 release) *and* the native irix C compiler.
That's on irix-6.5.14m. gmake version 3.76.1.

I looked at the rest of the thread and think that before starting
mucking around with the code or the system header files maybe you
should ask yourself 'and if it was a compiler problem?'

just my 2EU
jf

> 
> 
> gcc -DHAVE_CONFIG_H -I. -I. -I../config -I./../regex-src -g -O2 -c 
> statfs.c -o statfs.o
> statfs.c: In function `get_fs_stats':
> statfs.c:138: too few arguments to function `statfs'
> statfs.c:144: structure has no member named `f_bavail'
> statfs.c:144: structure has no member named `f_bavail'
> *** Error code 1 (bu21)
> *** Error code 1 (bu21)
> 
> 
> Thanks for any feedback,
> 
> Eric Zylstra

-- 
Mix a little foolishness with your serious plans; it's lovely to be
silly at the right moment. -- Horace

Re: your mail

[Joshua Baker-LePain]

On Tue, 19 Feb 2002 at 10:08am, Monserrat Seisdedos Nuñez wrote

> two questions:
> * i have a linux box doing amanda backups (amanda:as user bin)
> to an aix box (amandad:as user root), when linux does the backup in aix,
> which user run the program in aix machine?. 

It runs as whatever user you told it to when you ran ./configure on the 
AIX box.  Except for tar backups, which get started by the setuid root 
runtar wrapper.

> * amanda runs tar in the aix machine, but i have aix's tar and
> gtar, which one it is running?, i would like to run gtar because i have a
> exclude file and with aix's tar it doesn't work. The first in the PATH is
> aix's tar, how can i tell amanda to use gtar without modifing the path?
> 
amanda is running whichever one it found at ./configure time.  This info 
should be in config.status (if you kept it) or /tmp/amanda/amandad*debug.  
You *must* use GNUtar for amanda.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Joshua Baker-LePain]

On Mon, 26 Nov 2001 at 8:39pm, Liam Marshall wrote

> How do I unsubscribe from this list?

By following the directions at www.amanda.org, which are similar to the 
directions for every other majordomo mailing list out there.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Stephen Carville]

On Wed, 10 Oct 2001, Joshua Baker-LePain wrote:

- On Wed, 10 Oct 2001 at 11:38am, Stephen Carville wrote
-
- > dumpcycle   7 days
- > runspercycle10
- > tapecycle   12 tapes
-
- > Backups are run Monday thru Friday. I figure this setup will give me a
- > full backup every Monday. Is that right?
-
- If backups are only run Mon-Fri (5 days), and your dumpcycle is 7 days (1
- week), then your runspercycle should be 5, not 10.  Runspercycle is how
- many runs per dumpcycle, not tapecycle.

OK, I think I've got it :-)

What is important is that if any tape -- say Wednesday's tape -- is
destroyed (earthquake, fire, airplane) I can still restore the
databases files up thru Tuesday from whatever tapes are in the
archive.  The DBA tells me he can restore the entire database from the
files on the volume I am backing up as long as I can restore these
files.  Because of how the tapes are rotated, that means all I can be
sure to have is Monday's and Tuesday's tapes plus all the tapes from
last week.  If the amanda scheduler makes this possible, then I am
happy.

- > Now how do I configure amanda to do differential backups (level 1) on
- > Tuesday thru Friday?  One email I found in the archives suggested
- > setting bumpdays to 4 but wouldn't that mean level 0 for four days?
-
- You're trying to force amanda into your schedule (full Monday,
- incremental otherwise).  The way amanda expects to be run is to be given
- the parameters above, and then be left alone to set her own schedule of
- full and incrementals within those parameters.  IOW, she'll try to balance
- out the backups so that each night's are about the same size as the
- previoius nights while making sure she does (in your case) level 0 of each
- FS at least once a week.

Like I said above, it doesn't matter as long as I can be certain that
I can restore the database and systems files.

-- 
-- Stephen Carville http://www.heronforge.net/~stephen/gnupgkey.txt
==
Government is like burning witches:  After years of burning young
women failed to solve any of society's problems, the solution was to
burn more young women.
==

Re: your mail

[Joshua Baker-LePain]

On Wed, 10 Oct 2001 at 11:38am, Stephen Carville wrote

> dumpcycle 7 days
> runspercycle  10
> tapecycle 12 tapes

> Backups are run Monday thru Friday. I figure this setup will give me a
> full backup every Monday. Is that right?

If backups are only run Mon-Fri (5 days), and your dumpcycle is 7 days (1
week), then your runspercycle should be 5, not 10.  Runspercycle is how
many runs per dumpcycle, not tapecycle.

> Now how do I configure amanda to do differential backups (level 1) on
> Tuesday thru Friday?  One email I found in the archives suggested
> setting bumpdays to 4 but wouldn't that mean level 0 for four days?

You're trying to force amanda into your schedule (full Monday,
incremental otherwise).  The way amanda expects to be run is to be given
the parameters above, and then be left alone to set her own schedule of
full and incrementals within those parameters.  IOW, she'll try to balance
out the backups so that each night's are about the same size as the
previoius nights while making sure she does (in your case) level 0 of each
FS at least once a week.

I haven't played around with bump settings, so I can't help there.  I can
just say that, unless there's a *really* compelling reason, it's best to
let amanda set her own schedule.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Joshua Baker-LePain]

On Thu, 23 Aug 2001 at 4:53pm, chandra wrote

> I have progressed from tape labeling to amcheck. At the moment I am using root
> as a backup user for testing just to get it right. I have done the entries for 
>services
> and xinetd.conf and restart xinetd. Now when I run amcheck I get the following error.

You should really test as the user you intend on running amanda as,
otherwise the testing really isn't valid.  What user did you tell
./configure when you compiled amanda?

> ERROR: localhost: [access as amanda not allowed from root@localhost]
> amandahostauth failed

You need a file called .amandahosts in the $HOME of the amanda user
(determined at configure time).  In that file you need an entry for
localhost and for 'localhost root' (the second only for amrecover).
Again, the amanda user.

> The reason I am using root as a testing is because I cannot get to do the test as 
>amanda
> su amanda "amcheck DailySet" comes up with error
>
The way to do that is:
su amanda -c "amcheck DailySet"

And please turn off Outlook's send as HTML "feature" -- email is a text
medium and HTML just wastes bandwidth.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[Bill Carlson]

On Thu, 21 Dec 2000, Mark Berger wrote:

> Hi,
>
>
> I  was inspired to mess with the Amanda program for the nth time recently.
>
> It seem to have the feature that I want  - & some that I do not want as
> well!
>
> What I cannot seem to figure out is supposing I have 8 machines  ( 1 main
> server & 7 others) & want to do a full backup of the main server every night
> & a full backup of one other machine per day either in rotation or ad-hoc
> how can I persuade amanda to allow me to label / rotate tapes manually &
> always do a full backup of the specified machine?

Look into amadmin  force. You could script this so that full dumps
are scheduled for when you want.

However, I recommend against this. Amanda's scheduler does an excellent
job of deciding when to do a full dump of each file system in the interest
of keeping the time for the amdump run roughly the same.
Forcing full dumps will just screw up the schedule and Amanda will do a
better job than what you are planning. $.02

>
> It appears to me that much effort has gone into the scheduling etc but I am
> completely unclear as to how I might just say "backup machine 10.0.78.56 (
> or whatever ) in full to whatever tape happens to be in the drive - no
> arguments - no wrong tape etc just do it!" . - Or for that matter restore a
> file / group of files from tape in drive to machine
> xx.xx.xx.xx:/directory/subdirectory

You seek amrecover, some time spent with the documentation would be well
spent.

Bill Carlson
-- 
Systems Programmer[EMAIL PROTECTED]|  Opinions are mine,
Virtual Hospital  http://www.vh.org/|  not my employer's.
University of Iowa Hospitals and Clinics|

Re: your mail

[Joshua Baker-LePain]

On Thu, 21 Dec 2000 at 2:34pm, Mark Berger wrote

> Hi,

Hi.  These types of questions really belong only on amanda-users.  So
that's where I've responded, and set follow-ups.

> I  was inspired to mess with the Amanda program for the nth time recently.

If you devote just a little more time to your "messing with", I bet you'll
be very pleased.

> What I cannot seem to figure out is supposing I have 8 machines  ( 1 main
> server & 7 others) & want to do a full backup of the main server every night
> & a full backup of one other machine per day either in rotation or ad-hoc
> how can I persuade amanda to allow me to label / rotate tapes manually &
> always do a full backup of the specified machine?

This is easy.  In your amanda.conf, you specify a dumpcycle of 7 days, and
runspercycle of 7 (assuming you're going to run amanda every day).  Your
tapecycle should be at least 8, but make it as long as you need a history
for.

Where you handle the "full of this machine every day, full of these others
once a week" is in the disklist.  For the machine you want a full backup
of everyday, you use, e.g., a dumptype like "always-full" which, in its
definition, includes "dumpcycle 0".  For the others, you use a different
dumptype, e.g., comp-user, with no dumpcycle definition.

This is exactly what I do.  I do full backups of all my machines' /var and
/ partitions every night, and rotate full backups of everybody's /home
drives.

> It appears to me that much effort has gone into the scheduling etc but I am
> completely unclear as to how I might just say "backup machine 10.0.78.56 (
> or whatever ) in full to whatever tape happens to be in the drive - no
> arguments - no wrong tape etc just do it!" . - Or for that matter restore a
> file / group of files from tape in drive to machine
> xx.xx.xx.xx:/directory/subdirectory

Restoring is handled by amrecover (for picking and choosing) and amrestore
(for getting an entire dump image), which both have very informative man
pages...

Hope this helps.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

Re: your mail

[David Lloyd]

Joi!

> Amcheck tests and reports on all of these conditions.  Didn't you put it
> into your nightly backup script as suggested in the install guide?

Yes. And amdump still dies with no error if one of them doesn't work.
Personally, I would be inclined to set the logic to:

* can't write to /tmp
 - tough
 - having looked at /tmp/amanda the stuff in there is NOT important at
all to an AMANDA backup that otherwise would function

* can't write to amandates
 - serious problem
 - do a level 0 dump of everything, email off how to MANUALLY fix this
problem

* can't write to dumpdates
 - same as amandates

> /tmp itself is supposed to be world read/write, even user nobody can
> write there under most unixes.  If /tmp isn't world writable all kinds
> of things break.  I wouldn't hold this one against Amanda...

The fact that it can die, without a trace and no error is rather
unnerving. As I said above, I consider the backups far more important
than the debug data in /tmp/amanda (call me strange, but amanda is free
but if I lose the data we lose thousands)

DL
-- 
Do you believe in life after love?
 I can feel something inside me, say
 I really don't think you're strong enough

Re: your mail

[Joi Ellis]

On Fri, 17 Nov 2000, David Lloyd wrote:

>It appears that AMANDA must write to:
>
>* its config directories
>* /etc/dumpdates (if you're using dump)
>* /etc/amandates
>* /usr/adm/amanda or wherever you've set the log directories to
>* /tmp
>
>If it can't write to any of them it can simply die silently. No error.
>Nothing - especially if /tmp fails. amrecover and the index server seem
>to be the most susceptible to non-writable directories, although amandad
>and almost anything else that logs to /tmp can fail silently.
>
>I'm inclined to hack the source and stop it from dying without a trace;
>maybe a syslog of messages.critical saying there is no AMANDA logs would
>

Amcheck tests and reports on all of these conditions.  Didn't you put it
into your nightly backup script as suggested in the install guide?

/tmp itself is supposed to be world read/write, even user nobody can
write there under most unixes.  If /tmp isn't world writable all kinds
of things break.  I wouldn't hold this one against Amanda...



-- 
Joi Ellis
[EMAIL PROTECTED], http://www.visi.com/~gyles19/

Re: your mail

[David Lloyd]

Joi et al:

> (I'm not positive about the read-write, but it definately needs read.)

It appears that AMANDA must write to:

* its config directories
* /etc/dumpdates (if you're using dump)
* /etc/amandates
* /usr/adm/amanda or wherever you've set the log directories to
* /tmp

If it can't write to any of them it can simply die silently. No error.
Nothing - especially if /tmp fails. amrecover and the index server seem
to be the most susceptible to non-writable directories, although amandad
and almost anything else that logs to /tmp can fail silently.

I'm inclined to hack the source and stop it from dying without a trace;
maybe a syslog of messages.critical saying there is no AMANDA logs would
be better than being stuck without a backup!

I must confess - although I've found AMANDA an excellent backup system
and I've restored from it a number of times - I find that its error
reporting is absolutely woeful and/or non-existent.

DL
-- 
Do you believe in life after love?
 I can feel something inside me, say
 I really don't think you're strong enough

Re: your mail

[Joi Ellis]

On Thu, 16 Nov 2000, Matt Glaves wrote:

>Date: Thu, 16 Nov 2000 10:37:08 -0500 (EST)
>From: Matt Glaves <[EMAIL PROTECTED]>
>To: [EMAIL PROTECTED]
>
>
>I have just inherited a system running Amanda.  I have performed a few
>test restores and it appears that the system is only backing up files
>which are world readable.

User backup should be added to the same group which owns the disks, 
and that group should have read or read-write access to the physical
devices.

(I'm not positive about the read-write, but it definately needs read.)
-- 
Joi Ellis
[EMAIL PROTECTED], http://www.visi.com/~gyles19/

Re: your mail

[Rob Simmons]

doh.  I'm sorry about that.

-- 
Robert Simmons
Systems Administrator
http://www.wlcg.com/