know for sure what Amanda
is going to want. Otherwise Amanda is going to refuse the tape and the
backups will "fail" (go into degraded mode and require a manual amflush
the next day).
One simple way would be to do this a few hours before the "real"
amdump run:
amadmin tape | Ma
uot;driver-idle" value to find out what
the holdup is. Then you'll probably have to delve into the rest of the
file, and possibly the driver code, to see what the various states mean
(it's not as hard as it sounds -- driver is pretty easy).
>Paul Yeatman
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
so
eventually you will get it all cleared out.
But if you have a single image in there (even if it's in multiple holding
disk chunks) that is larger than what you can fit on a tape (apparently
3500 MBytes), you're stuck. Amanda cannot deal with images larger than
a tape (yet).
>
e right direction for how to do this?
See:
docs/RESTORE
http://www.backupcentral.com/amanda.html
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
(make
sure you "make distclean" or "rm config.cache" first). That will cause
the FSF's to be replaced by read loops.
>Henk
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
e that if -f is not on the command line.
>Peter
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
on't have time at
the moment to tackle that.
>Anthony Valentine
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
t:
ftp://gandalf.cc.purdue.edu/pub/amanda/gtartest-exclude
I tried it with your above example and it worked using "**", so
I suspect your problem is trying to use two patterns in the disklist.
Once you switch to putting the patterns in a file on the client I think
it will start
>... inetd on the client machine starts
>to eat more and more cpu ressources up to 98% ...
If it's inetd chewing up the resources, then you need to talk to your
OS vendor. It's unlikely Amanda can cause that.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
recall,
it was a problem in their changer configuration file.
Look in your amanda.conf for "changerfile". There should be a file
by that name with ".conf" on the end. What's in it?
>Sandra
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
d my previous letter
described (although possibly too briefly) what to do with it.
If you have trouble getting it going, don't hesitate to post.
>Rob
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
d to changer-src and run "make"? Does it get
created? How about "make chg-scsi"?
>Rob
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
chg-scsi pieces into your "normal" build tree.
>Rob
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
if it helped (it will at least log better
messages about what's going on). Any chance you could do that on the
server and this particular client?
Did you specify any kind of portranges during ./configure? What kind of
protection is sitting between the server and client, i.e. TCP wrappers,
a f
rceforge.net/fom-serve/cache/16.html
http://amanda.sourceforge.net/fom-serve/cache/140.html
And, as Olivier says, you should use "amcheck" to test out your
configuration. When it's happy, amdump will have a much better chance
of working.
>travis
John R. Jackson, Technic
oving it when done might be safer.
As I said, we don't know of any security problems in Amanda. Whoever was
poking at me yesterday may have just been playing around. But if someone
has found a hole and we haven't heard about it yet, you should do what
you can to protect yourself (good advice in any case).
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
/amanda/chg-zd-mtx -slot next
>taper: slot /usr/bin/mt:: no tape online
That's definitely not a good response. My guess is that text ended up
on stdout or stderr by mistake and taper soaked it up as the answer.
You should probably look at /tmp/amanda/changer.debug.drive*. That should
show you what was going on inside chg-zd-mtx.
>Peter
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
oing
to take at least two hours (maybe even more) to do that file system.
>Edward.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
along with this file system (if you used
--with-pid-debug-files on ./configure or are running 2.4.2p2) and see
what kinds of timestamps it shows. Take a look at the last modification
time on the file as well (ls -l).
>Edward.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
l debugging to track down exactly where it was at.
>jobst
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
;by hand"), you can use amrestore:
mt rewind
amrestore -p $TAPE no-such-host > /dev/null
This will search the tape for a client named "no-such-host" and, of
course, not find it. But along the way it will report all the images
it skips.
>Luc Lalonde
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ut "configurations".
What, exactly, did it say that led you to think it was looking in
/usr/local/etc/amanda/dailies?
>Chris
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
N file). All the other Amanda statistics are based on what was
*successfully* done. So if the image being processed at the time of
the error was large, they could be way off from where the error happened.
>Drew
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
on't ever have an employment problem as long as there
are cars (or snake oil :-).
>Drew
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
clients
you want to test amrecover on, but the others should also (eventually)
be rebuilt to make sure the change does not affect them when going
through your firewall/NAT.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
stream_client.diff
Oh, yeah. That problem. :-)
Go to the Amanda patches page (www.amanda.org/patches.html) and this
very problem is listed.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
on (it will look up "/" by default).
>Rob
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ss someone has a problem with it, is pretty
simple and won't take long to implement once I get a few free minutes.
>Michael Fritsch
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
uot;
file system based on what happens to be running at any given moment.
As far as I know, it doesn't contain any real data that would be useful
on a restore (but that could easily depend on your OS).
>Rob
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
it determined /sbin/tar is not that,
so it will not use it.
>ltconfig: you must specify a host type if you use `--no-verify'
I think this is another symptom of the above config.guess/config.subs
problem.
Another possible workaround would be --disable-libtool on the Amanda
./confi
hat will it want as
an argument? That's what you should put in the disklist.
>Edwin R. Rivera
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
prepended. So if this is entered:
exclude list ".amanda.excludes"
the actual file passed to GNU tar would be
/var/.amanda.excludesfora backup of /var,
/usr/local/.amanda.excludes for a backup of /usr/local,
and s
>Can I upgrade seamlessly to 2.4.1 to 2.4.2 ...
Yes. There should not be any compatibility problems between those two.
>Ben
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
s is different than on AIX because Solaris lets me "su" to it
with one group of settings and AIX insists on different settings.
> Brian
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
out).
If it's really just too much data all of the sudden for some reason,
you'll have to go to plan B and add tape capacity in some way (longer
dumpcycle, larger tapes, more tapes per run, multiple configurations,
etc).
>Oscar
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
/amanda/gtartest-exclude
which is what I use whenever I have to figure out a pattern (and you're
right -- they are hard to deal with).
I think the GNU tar info document also contains information on the
patterns.
>Sheldon Knight
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
efix and put the config
directories in /var/amanda using --with-configdir and it works fine.
Are you sure you removed config.cache (or ran "make distclean") before
trying --with-configdir?
>Dan
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
es to emulate a tape. You could then
take that area and write it to a CD.
The amanda(8) man page in that branch has been updated to document this.
>--Thomas Kleymann
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
calls it), that reads from the raw disk devices so the Amanda
user just needs read access to that (*).
If you use GNU tar, Amanda runs it under a program called runtar that
it provides, and runtar is setuid root.
>Rob
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
(*)
or equal to 1024. You must set --with-udpportrange (the
UDP ports) to values less than 1024.
In addition, you'll need to set up your firewall/NAT to let both ranges
of ports through as is.
However, there is a bug in 2.4.2p2 that will cause amrecover to fail
in the above setup. I'm w
in the
/tmp/amanda/sendbackup*debug file on this client for that disk?
>Anthony Valentine
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
uld
also be interesting to look on the client and see if the backup is still
trying to run.
>Sandra
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
lt) and run it on the
GNU tar. That should tell you what it has open.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
P.S. Please turn off "send as HTML" in your mailer. It's just a waste
of bandwidth.
in
your amanda.conf.
If that does not turn out to be it, I sent you a patch I recently posted
to amanda-hackers to get around the 64 KByte UDP packet size. You might
give it a try.
Make sure you use the "third try" version :-), as I've posted it more
than once.
> Enri
uld never, ever, run any
Amanda command (except amrecover) as root. They can leave files laying
around with the wrong ownership.
>Frank
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
posted another (untested) idea that might apply, at least
in part.
>Christopher Masto
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
. In particular, look at the first
and last lines and see how long the estimates took. You may need to
crank up etimeout in amanda.conf.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
they
seem to have good data in them (should be paths and file names)?
On the tape server, what's in /tmp/amanda/amindexd*debug?
What happens if you run "amadamin gilsdorf find localhost /etc"?
>Frank
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
es (indexdir) and tapelist (same directory as amanda.conf or
"tapelist" in amanda.conf) separate between the two configurations.
>Thanks again for additional help.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
AM (64*1024) to MAX_DGRAM(512*1024) ...
FYI, I checked with my network wizard and the upper bound for an IP
packet is 64K. There is some protocol involved, so the actual upper
limit for the UDP packet is slightly smaller.
> Enrique Rodríguez Lázaro
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
...
That's good to hear (well, it's good you may have found the problem :-).
>Sean Noonan
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
em with pipes and GNU tar, or maybe it was the network, that caused
the whole process to hang. Amanda doesn't do anything terribly magic
with either, so my guess was a MacOS problem.
I have a world class Mac wizard literally next door, but we haven't had
time to sit down and work o
to zero. If you do an "amadmin
tape" before the run, slot zero means the first tape, slot 1
means the second and so on.
>Luc Lalonde
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
setting the modes
back to 550 but without setuid.
> Brian
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
nt. So my
guess is you need to whine loudly at MicroSoft (which will do you no
good whatsoever) and also ask the Samba folks about this.
>Sean Noonan
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
he client and run:
df /usr/local/libexec/rundump
file /usr/local/libexec/rundump
(run them as the Amanda user). The first will confirm the file is coming
from where you think it should. The second should tell you about the
architecture it was built for.
>
>... I get this strange warning in the logfile:
>...
>? SSL: Error error setting CA cert locations: error:::lib(0)
>:func(0) :reason(0)
>? trying default locations.
Those messages come from smbclient. You'll have to ask the Samba
folks what it means.
>Jens
If you want compression, use (e.g.) /dev/rmt/0mn. If you don't, use
/dev/rmt/0ln.
>- Sam
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
e requests? ...
Not me. Anyone on this list should be used to a few hundred letters
a day :-). And anything that makes Amanda better, such as getting the
SCSI changer going on AIX, is a good thing (IMO).
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
have a 2 GByte individual
file size limitation). But the whole image still has to fit before
it is sent to tape. If you don't have enough holding disk to do that,
the image goes direct to tape.
Some future work will change that, but that's the way it is for now.
>
midxtape stream tcp nowait backup /opt/amanda/libexec/amidxtaped amidxtaped
Without that trailing "amidxtaped", argv[0] is NULL and that's why the
dbprintf failed.
>\Ben
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
library -lgen: not found"
stuff for Amanda. I looked at this a little more. Both libgen.a and
libtermcap.a are in /usr/ccs/lib, so that may be yet another chunk the
previous person removed (or never installed).
If you get truly tied up in knots trying to make this machine into
somethi
arious files and stuff. If that's
>the case shouldn't I be 'make'ing on the 2.6 box (named taipei) instead?
I'm sure with enough work and experience you could eventually make it
happy :-), but at some point it's not worth the hassle.
>Chris :)
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
o be working. I know the list went through some ISP changes recently,
and things have been a bit shaky since then, but I'm pretty sure the
administrator is working on it. I'll forward your note just to be sure
he's aware.
>Toomas
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>Sooo, no go on that road.
Not too surprising.
>Btw, I *did* manage to get a copy of 'flex' available to the 2.6 box (via
>NFS) and it wasn't able to build.
>
>Are either of you (Olivier/JRJ) interested in the output?
Sure. Maybe it's something minor. Go
ot;/" mounted from /dev/dsk/c0t3d0s0?
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ot needed all that often any more. Most Amanda problems can
be diagnosed from the logs and don't need a program debugger. Which is
OK since most people don't know (and, because of the stability of modern
systems, don't need to know) how to run them.
>Olivier
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
y" estimates, but I wouldn't bet on them
being a whole lot faster. So it seems to me an etimeout of around 1800
would not be unreasonable for your configuration.
The amandad*debug output more or less matches this problem. Amandad got
a duplicate request, which was planner getting tired
have an old version of Amanda
being run from inetd?
>Don Mohlmaster
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
rver). It doesn't seem
worth the few KBytes of binaries it saves, and doing the same thing on
every machine is much simpler.
Every machine that gets backed up needs the client pieces, even the
server. So the server should get everything in any case.
>Kevin
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
'm not sure datagram packets can be that big. You should probably put
that back the way it was.
> Enrique Rodríguez Lázaro
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
env CFLAGS=-g ./configure ...
* Use --disabled-shared.
You need to rebuild everything, not just amidxtaped, because of the
Amanda libraries it brings in. But you only need to install/test the
resulting amidxtaped binary. With --disable-shared involved, it should
be self-contained.
&
>I get always the Message:
>[disk /dev/dsk/c0t3d0s0 offline on arane.webcom.de?]
Did you run amcheck? What did it say?
What does /tmp/amanda/sendsize*debug on arane.webcom.de have to say
about the "failing" disk?
>Bernd Zimmermann
John R. Jackson, Technical Software
you with.
It appears to be a general changer interface routine that is not yet
supplied for AIX.
>Anthony Valentine
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
> server/dev/sda3 lev 0 FAILED [Request to server timed out.]
What are the **entire** contents of /tmp/amanda/sendsize*debug on "server"
that correspond to one of the failures?
My guess is "etimeout" is too small in amanda.conf.
John R. Jackson, Technical Soft
e hunts for).
>|amv: /home/amv/amanda-2.5.0 > make
>...
>"Makefile", line 449: make: 1254-055 Dependency line needs colon or double
>colon operator.
You **must** use GNU make to build from the source tree (don't ask me
why -- it's a deep dark GNU mystery).
>Anthony Valentine
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
e). That's controlled with maxdumps. So changing
inparallel will not have any effect, unless the client that's crashing
is the taper server itself.
Is maxdumps set higher than one for this client (based on the dumptype)?
You can tell with "amadmin disk ". If so,
set it back to one
changer.conf
>
>firstslot=0 1st tape slot
>lastslot=7 Last tape slot
>cleanslot=7 Slot with cleaner tape
This probably isn't related to your current problem, but I don't think
you want lastslot to be 7. It should be the last slot that has a "real"
tape in it. Slot 7 (apparently) has your cleaning tape, so lastslot
should (I think) be 6.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
getting
big enough as it is :-), but keep it in the back of your mind.
>Chris
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
al/libexec/chg-scsi.
I wouldn't recommend running all of 2.5. It's **highly** experimental.
But all of the active chg-scsi work is being done there, and it should
be stable, so that's the best way to work with that particular piece.
>Anthony
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
k up everything
changed/new since the full dump. If your data is very active,
that could use up your holding disk faster than you might expect.
* Recovery should be fine. Amrestore/amrecover both know how to
read from the holding disk.
>Pam
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
the
contents, or rename it yourself.
>I am still doing tests with my installation of amanda.
>
>I use amanda build: VERSION="Amanda-2.4.1p1" on Debian potato.
Ummm, if you're just getting started, you might want to get a new
Amanda release. 2.4.1p1 is years old. 2.
and level, file system type, amount and kind of activity,
current number of sunspots, ... :-).
>Denise
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Amanda starts writing, *all* the previous images on that tape are gone.
So if something goes wrong during the run and the dump is not done,
and then the disk fails, you don't have a usable backup.
>Pam
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
;ll just have to
play with this and see how it goes.
I'm also not sure how amrecover will behave toward "no-reuse" tapes.
It should be pretty easy to write a script that, given a set number to
bring online, does the appropriate amadmin calls by looking through the
tapelist file for all the available labels. Ones that "match" get set
to "reuse", all others get set to "no-reuse".
>Olivier
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
me program gave the kernel a buffer that
was smaller than the record size reported by the drive (32 KBytes).
Amanda is extremely unlikely to have done that. My first suspicion
would be a kernel or driver buffer management problem. Maybe you could
ask those folks?
>Toomas
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
$volume ]
then
find $volume -type c \
-exec $DEBUG chgrp backup {} \; \
-exec $DEBUG chmod 0640 {} \;
fi
done
>Dave Warchol
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
till fail.
At least part of the difficulty with all this is the differences
between OS's. For instance, you mentioned setting the user and group
in inetd.conf. None of my OS's support that, so I wouldn't have had
any idea to advise you to look for that type of problem.
>Josh Kuper
>As it wasn't defined the spec of a EXB-8505 with 160 meters cartridge
>in the amanda.conf, so I have run tapetype. ...
Thanks. Would you mind posting this to the FAQ at www.amanda.org, please?
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Then divide that by the
number of disks. Fudge that up quite a bit and see if setting etimeout
to that makes this work.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
".
Either do it by hand, or set up a cron job. You might decide whether
now is the time to change "record" to "yes" or not.
>I would like someone to mentor me through the process and let me ask a few
>questions to get started.
I think you'll find this mailing list a very easy going place to get help,
so post away :-).
And welcome to Amanda!
>John Holstein
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ging patches to amindexd to track what it's doing inside.
Are you building from source so we can do that? You might want to build
with "-g" so if we need to fire up a debugger to check something out,
it will be ready and willing.
>Dan
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
me is now part of the amandapass file. See the docs/SAMBA
file. It's been updated to show the new syntax.
>David
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
eady graciously posted the info in an FAQ item, so I don't think he'll
mind the letter being posted). You might look in the E-mail archives
for Subject "NAT" -- that was the other thread. Maybe they will help.
>Ben
John R. Jackson, Technical Software Specialist, [EMAIL P
er, but it seems to give a rough
idea of how much output has been generated in more or less real time.
> Erick Bodine
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
xx.c
ationship :-), let me know
offline (there's no sense annoying the rest of the list with our chatter)
and I'll put together a first cut at logging what's going on.
>Dan
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
x27;ll not only help yourself next time you run into the same thing,
you'll also be helping other people.
>Josh Kuperman
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
to check it because a lot of people were
>waiting after the server to be up....
We've all been in that situation :-).
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ot work, it's going
to take some more head scratching (and probably some patches to try and
log what's going on -- are you up for that?).
>PS. the message is in plain text. ;-) ...
Thanks! :-)
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>I'm having a problem backing up files on a NT-machine ...
As Olivier said, all those errors are from Samba, not Amanda. You need
to ask your question on one of their mailing lists.
>Jens Rohde
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
801 - 900 of 2203 matches
Mail list logo
