odd" strategy options
turned on)?
If nothing pops to mind, you should be able to "amadmin force"
those two disks and get Amanda to rebuild the level 0 files on the
next run.
>Oscar
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ve happened since
plain 2.4.2.
However, take a look at the advfs patch on www.amanda.org. It's needed
for Linux.
>Shawn
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
for client2 that you posted says the former (krb4). My guess
is the output for client1 says the latter (bsd). It looks like you need
"-krb4" on your amandad line in inetd.conf to get krb4 security turned on
(check client2 and see how it is started).
>--Ruth Anne
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
hip
or modes.
* Some other process (e.g. a cron job) is going around cleaning out
"old" files in /tmp. You will either need to tell it to leave
/tmp/amanda alone, or put the Amanda temp area someplace else (I
change mine using --with-tmpdir=/var/amanda/tmp).
>Sascha
John R
the most common errors to happen when you
start with Amanda. There are several debugging steps in the Amanda
FAQ at www.amanda.org. Give those a try and if you still have trouble,
please post back exactly what you tried and what the results were so we
can help you get it working.
>Fabio d
e server is running everything. It initiates all
backups by reaching out to the clients (and yes, that client/server
terminology is confusing). The tape server does not just sit there
waiting for clients to connect and start dumping data at it.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
really means
is the test failed for some other reason.
Not saying that's what's going on with your setup, but it's happened
more than once.
>-J
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
oftware bucket. It has "space"
(netusage) that is allocated and deallocated as backups start and end,
and that space is used as a limiting factor for startup decisions,
but it does not represent anything real.
I recommend "logical" interface names, such as "fddi-back-
>Yeah, still have a problem. ...
Nuts! I was hoping your problem would just go away on its own :-).
>Am I doing this on the troublesome client or on the tape host?
The client.
> [EMAIL PROTECTED]
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
y ones with any intelligence. In other words,
"neat" things like this take effort, sometimes magnitudes more than you
might think just from the basic idea.
That's not a gripe or anything. I've certainly made my fair share of
"little suggestions" to others, even t
checking for readline in -lreadline... yes
If that doesn't work, you're back to looking at config.log to find out
why the test programs failed.
>--with-debug-days=1
Any particular reason you're lowering this value? I'd think you'd want
at least a weekend worth of logs left around.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
e this:
with_user=amanda
with_group=amanda
with_tcpportrange=2000,2040
Just keep this file around and you'll know what you used the last time.
>Jeremy Wadsack
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
#x27; in the name means "Beta", right? As in,
we're still coding (and documenting), and if you run into trouble,
we sort of expect you to be able to do some debugging.
If you want a more stable release, I'd recommend 2.4.2p2.
>--david
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
gure so Amanda uses the script:
make distclean
GZIP=/path/to/the/test/script ./configure ...
# Make sure ./configure found the "right" gzip
make
su -c "make install"
Then after a failed run, see if any of the /tmp/gzip.log.* files show
anything interesting.
> [
cc FAQ
someplace), but be that as it may ...
>Now i have another problem, when i make the source i get the next error:
>
>collect2: ../common-src/libamanda.a:not a COFF file
Here's what I would try:
make distclean
./configure --disable-libtool ...
make
John R. Jackson, Technical
ull dumps and set tapedev to "/no/such/device" to force amdump to
run but leave everything in the holding disk. Then some clever hard
linking (to prevent removal) and multiple amflush runs should be able
to accomplish what you want.
>Frank
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
member named `ces_data'
>...
As long as you don't need that changer (which you probably don't)
then I'd just go into config/config.h and #undef'ing HAVE_CHIO_H and
HAVE_SYS_CHIO_H. That should effectively disable that part of the build.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>... Can Amanda backup the 4 filesystems to the 4 tape-drives
>simultaneously from one config file? ...
No. That will take a major overhaul (although it's definitely needed).
>Aaron Rainwater
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
no difference as far as Amanda is concerned. It still makes a
network connection (even though it's through the local host interface).
In particular this means inetd has to be set up right, which is the most
common source of trouble.
>Don Potter
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
27;t have
a cleaning tape, set it to something outside the first .. last bounds
(e.g. "cleanslot=99").
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
look "new" to Amanda and so full dumps
will be required to get started unless you mess around with the database.
>Terri Eads
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>I wonder how hard it would be to train amanda to rotate holding
>areas (partitions) on a raid instead of tapes? ...
That's already part of 2.4.3 (or the amanda-242-tapeio CVS branch).
>Gene
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
tools (mt, dd and whatever
restore program matches the dump format).
The one thing you might want to be careful about is compression. Just to
be safe, I would probably *never* software compress my OS file systems
on the off chance I didn't have the uncompress program during such a
disaster recovery.
>S.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
to make any further guesses as to where it's going wrong.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
..
You can't have a user without a home directory. The home may be shared
with other users, but it can't not exist (well, I guess it can, but it
is still defined in /etc/passwd even if the directory does not exist).
The home directory will be required whether you use .amandahosts or
.rhosts. The only thing different is which file is referenced.
>jfk
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
appened in config.log.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>here's a patch for Amanda 2.4.2p2 to ignore xfsdump's occasionally
>output on busy filesystems ...
Thanks.
Before I put this in, is there any particular reason you're adding
DMP_WARNING? Why not just add these entries as DMP_NORMAL?
John R. Jackson, Technical Softwa
be hard to add
to the framework that's there.
Have you actually tried a "du -sk" to see if it's any faster? Both it
and tar have to traverse the whole directory tree and stat everything.
If you're doing whole file systems, "df -k" would be the way to go.
Or sw
newer versions of Amanda. Earlier ones hard coded
the Samba login name. So there may be a mismatch between what you have
in the file and what the code can deal with.
If you're not already doing so, any chance of upgrading to 2.4.2p2?
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
or not.
So, does that explain things a bit?
If you have other questions, just post.
>Stewart Dean
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Did you do something
special to change that?
Prior to building amandad, make should have built and run ./genversion.
I assume, since you didn't say anything, that that went OK, i.e. you
didn't get load errors and it ran?
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
uld need to see the corresponding amdump. file that goes along
with one of these errors, and possibly all the client logs for that same
run. If you want to contact me offline of the list for this, that's OK.
In the meantime, you might try dropping maxdumps back to 1. That may
or may not he
er trying
amcheck? If it's not getting updated, that also means it's probably
the inetd child failing before it can exec the real program.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>it only croaks on the amcheck...
Huh??? It **doesn't** fail for amdump?
>the amandad will Hangup before providing any output..
Huh???
Did you try either of the other "debug tricks" I sent?
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
to folks
who have to work across firewalls, et al. Comments and suggestions
would be welcomed.
>-Kevin Zembower
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
dgram udp waitbackup /bin/truss amandad -fo
/tmp/amandad.truss /opt/amanda/libexec/amandad
Note that the user name ("backup" vs. "amanda") and paths to the binaries
on your system will be different.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ike?
>Don Potter
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ght just ignore setting
comprate since 50% is not all the different than 40%.
>may I run amdump 2 or 3 times the same day, so that all partitions get a
>level 0 dump ...
Yes.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
g from a previous failure.
* Verify that the amandad service in inetd.conf/xinetd is set to
"wait" (the default is "nowait").
* Make sure you don't have multiple cron jobs running at the same time.
>Sascha
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
majority of your data. If that single disk fails,
you're screwed.
I'm not disagreeing with the idea of using disk for backup. Just saying
it takes some thought to make it a safe setup (tape does too, of course).
>Jeremy Wadsack
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
position the tape myself to do a restore (rewind followed
by an appropriate fsf) because it's much faster on the types of drives
I have.
>> --a
> [EMAIL PROTECTED]
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ll
misbehave with easy tests :-).
>... Switched to the other IDE controller, still get the same behavior.
That leaves the cable, the drive itself, or tremendously bad luck
picking several tapes that all fail (my money is on the drive, or maybe
the cable).
>Steve Stanners
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
iting up something for the FAQ), but suggest
you check out the mailing list archives for the details.
Basically, both dump and tar do both good and bad things. You just
decide which you're comfortable with.
>Dick
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
e level (or
dropping back to do a full would be OK), then you should be in good shape.
If it gets most of them right and just a couple bump anyway, use "force"
on them for the run after next to get them to do a full.
>Ben
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
/Work" part
is forgotten). If you start amrecover from directory /home/abc, then
the file will be restored into /home/abc/project/xxx.c.
The moral of the store is, if you want to restore things back to their
original location, cd to the disklist entry name (e.g. /mnt/share/Work)
before start
won't, figure out what you
need and adjust the procedures until you're covered.
>Joshua Baker-LePain
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
put in libexec.
While you're at it, see if he'll send a copy to me and I'll put it in
the Amanda release.
If none of this works out (Nick may not be able to help for any number of
reasons), let me know offline and I can probably put something together
pretty easily. It looks like your "robot" command will be simple to use.
>JF
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
if it can be improved.
>Adam Haberlach
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
all be cleaned up some to make it easier to do.
It comes up often enough that it should be better supported.
>Raphael Becker
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
rwater
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
's a few billion dollars of worth floating around there,
and I only looked at the U.S. .com entries. There are almost 500
international entries and another hundred .edu's (and if you don't think
universities are in it for the money ... :-).
>-Kevin Zembower
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
cess to the devices
(and /etc/dumpdates) by group membership, which is perfectly reasonable,
but the xinetd folks screwed you up.
>-Aaron
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
re getting).
It's possible to patch this up, but by far the easiest way to fix it is
to do the install again (as root).
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ou need
multiple tapes to do the restore, which is common, remove and reset the
link whenever amrecover asks for a new tape.
>Jeremy Wadsack
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Upon even further review :-), it looks like you should be able to put
the device name on the command line with just "-d file:/whatever".
The place you'll have trouble is with the settape command when inside
amrecover, and in that case should use "settape server:file:/whatever".
I think. I'm just
Never mind. I see where it's happening now.
Try "-d your-tape-server:file:/home/amanda/dumps/".
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
27;t see
it parsing that (although I could easily have missed it).
>Jeremy Wadsack
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
een "checking for gtar ..." in the ./configure output.
What did it say on that line (and maybe the next couple of lines)?
BTW, I just tried this on my system and it "did the right thing".
We just have to figure out what is different about your setup.
>-mh.
John R. Jackso
latest version available on
>any of the GNU mirrors. If I need a newer version can you tell me
>where to find it?
It's at alpha.gnu.org. No, we have **no** idea why the GNU folks are
still offering that bad version on the normal mirrors. It's not very
nice of them.
>Jeremy Wad
t to in either the global section of amanda.conf or
the individual dumptype? What does "amadmin disklist "
say about maxdumps?
>Chris
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
emote
host ("server" or whatever the error message has been saying) then it will
trace through the sequence of steps used to validate that access attempt.
Hopefully something in that output will point out what's wrong.
>Jenn
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
".
Any core files in /tmp/amanda? If so, could you get a debugger traceback
and post it?
>Brian
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
" is an invalid directory
What do you get if you do this:
cd /var/amanda/daily/index/10.0.0.17/hda1
/usr/bin/gzip -dc 20020121_1.gz | sort | head
>Jeremy Wadsack
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
n mechanism just won't work? ...
Unlikely. The code to use .amandahosts is completely built in to
Amanda itself. It's "our" code so we can make it work.
This is a fairly common startup problem, so don't despair. It's almost
always a permissions, ownership or path problem of some type.
>Jenn
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
7;re all set.
If it does not, you're going to have to force the issue on each mount
via the front panel until all the tapes get rewritten once (sigh).
>As for the tape mounting at 35.0 gig
>density, I did not consider any change in it since Amanda was mounting
>the tape (again, the
te).
Consider yourself blessed :-). The estimates on some of my systems run
almost two hours.
>Sascha
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
cally /etc/fstab or
something similar). My guess is you don't have that set up properly
for "hda2" but it is correct for "hda1".
The easy answer to your problem is to use the mount point name in your
disklist instead of the disk name. In other words, change "hda1"
term :-). If you
have enough hardware, try moving the drive to a different controller,
swap the cables, etc.
>Steve Stanners
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
but it's harder than you might
think (it used to be a non-standard system call, so portability is
a problem).
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
P: Warning: unable to translate LABEL=/boot
Note that it's not sendsize whining about "LABEL=", it's dump itself.
My guess is your version of dump needs to be updated, and it wouldn't
surprise me to hear that that also fixes the permission denied problem.
>Dick
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
s entry with the name of the program. But passing that
as an arg to Amanda services really confuses them.
In any case, you don't have that problem, so this is all academic.
>Kurt L Vanderwater
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ough... latest I could find on ftp site
>ended in 1999.
Look toward the bottom of the Amanda web page (www.amanda.org). It refers
you to the searchable E*groups (Yahoo) mailing list archive.
>Kurt L Vanderwater
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>There is no time estimation done:
>...
>Estimate Time (hrs:min)0:00
What version of Amanda?
In the corresponding log.MMDD.NN file, what does the "STATS driver
startup time" line say?
>Sascha Wuestemann
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ted at 35.0 GByte density?
>The only thing different from the tapetype suggestion is that we
>changed the filemark from 0 to 2024 kbytes.
As above, that would only affect how much Amanda had to work with to
do the estimates. It would have nothing to do with actually writing to
the tape.
&g
the "reserve" parameter to a value less
than the default of 100% to allow Amanda to go ahead and do level 0
dumps into the holding disk (assuming they will fit) even when a tape
error happens.
>Michael Richardson
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
rising. There was some
discussion of this recently in this group so you might take a look at
the archives.
>I also found when I have more than one disk image dumped to the tape,
>I need to mt fsf to the correct disk file before running amrecover.
>What's up with that?
Same as above
got the URL right this time :-).
If you're doing Oracle backups, you might take a look at the dbbackup.*
files in that same location. They are what we use, although they are run
before running amdump rather than on the client at the time that "disk"
is done.
>Aaron Smith
John
ke sure you do a "make
distclean" (or "rm config.cache") before so old build information is
not left laying around.
>Do I have to use the
>same user on both server and client if I'm using gnutar?
No. The only rule is that you have to run Amanda on a machine (cl
ed from the Amanda web page) for info on the OnStream drives.
They have kind of a bad reputation here.
>Dick
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ives for "SERVICENAME" or ask me offline and I'll
send you the patch (it's not very large).
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
" limit? If so, you can use smitty to
change that.
>-edwin
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
do to solve the Out of Memory problem?
Add more memory? :-)
Or, more seriously, add more swap space.
>Is it related to the out of tape problem in any way?
I doubt it.
>-Kevin Zembower
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
that be a more sensible setup?
The general suggestion is that you have at least twice as many tapes
as your dumpcycle. That way, even if the most recent level 0 is bad,
you have at least one more you can fall back on.
>Reidar
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ss OK
DATE 2002-01-17
200 Working date set to 2002-01-17.
QUIT
200 Good bye.
The lines that do *not* start with a number (e.g. "SECURITY USER root")
are what I typed.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
-user to "backup" or start running amandad as "operator"
on that client.
> [EMAIL PROTECTED]
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
first file (32 KByte blocks) of each tape going out onto the
corresponding tape coming in to make them fit back into the normal cycle.
That's easier than rerunning amlabel on them which can re-order the tape
cycle and generally confuse things.
>Bryan Tonnet
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
st.
Try it and see. Take an Amanda tape, skip out a bunch of files then
do an "mt offline" on the no-rewind device. Look at the tape when it
comes out. All the media will be to one side.
>stephen.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ok like? Or, what's in amidxtaped*debug
on the client when you run amrecover?
>Robert Dale
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
to know what to tell the provider to do.
Tell them to buy a new tape drive that works. If they squawk about price,
reminde them how much time and money have been wasted on what they have,
and what losses have happened (or will) because backups are not working.
> Radu Filip
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
o the number 4 (at least as far as we care) other than
it's not zero (success).
>-Aaron
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
ch bigger (and separate) change for another time.
Please confirm this patch fixes the problem you ran into so I can get
it into the source tree.
>Volker Apelt
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
reporter.diff
re you using Linux? And does the current working directory file system
have the 2 GByte file size limit?
You probable need to do the restore though a pipe and not put it to disk
first, something like this:
amrestore -p /dev/whatever gateway hdb1 | /the/restore/program ...
>Len Pikulski
the disk is failing. Does errpt have
anything interesting to say? What happens if you dd the disk to /dev/null
(dd if=/dev/ru02 of=/dev/null bs=1024k)?
>-Aaron
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
27;s the program
that does not support exclusion lists, and you'll have to ask the Samba
team for help with that.
>Luc
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
solve your problem. At some point (don't recall
when) the default became --with-amandahosts. It might be that was
after 2.4.1p1.
>Jeremy Wadsack
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
e-device does is set a default for the amrecover tool.
You can override it (and the tape server) on the command line.
> Lalo
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
swer the remote host questions
with the exact same text as was in the message. It should trace the
actions in more detail and hopefully give some more information.
>Jeremy Wadsack
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
en around
to it. If you want the patch, let me know and I'll dig it up. Or you
could get the latest (-ramanda-242-branch) CVS code and work from that.
>-Ben Bullock
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
of course).
Do you have a /tmp/amanda/amidxtaped*debug file on the server? Is there
any indicatation amidxtaped started? Do any of the log files indicate
a problem starting it?
>Alexei
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
; loop terminated
and the "direct to tape" final pass kicked in, which caused the messages
you saw. If the dumpers had behaved, this would not have happened.
Just more symptoms of the one common problem.
>Mark
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
401 - 500 of 2203 matches
Mail list logo