eof problem with amrecover
I keep getting the following error when I try to recover a file with amrecover: tcsh /usr/local/sbin/amrecover -C WeeklySet1 -d /dev/nst1 -s redbaron.physics.s c.edu AMRECOVER Version 2.4.2p1. Contacting server on redbaron.physics.sc.edu ... 220 redbaron AMANDA index server (2.4.2) ready. 200 Access OK Setting restore date to today (2001-03-16) 200 Working date set to 2001-03-16. 200 Config set to WeeklySet1. 501 No index records for host: beauty.psc.sc.edu. Invalid? Trying beauty.psc.sc.edu ... 501 No index records for host: beauty.psc.sc.edu. Invalid? Trying beauty ... 200 Dump host set to beauty. Can't determine disk and mount point from $CWD amrecover setdisk /var/spool/mail Warning: no log files found for tape WeeklySet111 written0-00-00 Warning: could not open holding dir /jekyl17/amanda: No such file or directory 200 Disk set to /var/spool/mail. amrecover ls 2001-03-15 . 2001-03-15 adamt 2001-03-15 amanda 2001-03-15 copty 2001-03-15 kothari 2001-03-15 nussinov 2001-03-15 singh 2001-03-15 sudheer 2001-03-15 yumiceva amrecover add adamt Added /adamt amrecover settape redbaron.physics.sc.edu:/dev/nst1 Using tape /dev/nst1 from server redbaron.physics.sc.edu. amrecover extract Extracting files using tape drive /dev/nst1 on host redbaron.physics.sc.edu. The following tapes are needed: WeeklySet112 Restoring files into directory /tmp/amanda Continue? [Y/n]: Y Load tape WeeklySet112 now Continue? [Y/n]: Y ./adamt tar: Unexpected EOF on archive file extract_list - child returned non-zero status: 3 Continue? [Y/n]: Y amrecover -- The amidxtape.debug file is -- amidxtaped: debug 1 pid 5970 ruid 422 euid 422 start time Fri Mar 16 11:32:12 20 01 amidxtaped: version 2.4.2 SECURITY USER root bsd security: remote host beauty.psc.sc.edu user root local user amanda amandahosts security check passed 6 amrestore_nargs=6 -p /dev/nst1 beauty ^/var/spool/mail$ 20010315 Ready to execv amrestore with: path = /usr/local/sbin/amrestore argv[0] = "amrestore" argv[1] = "-h" argv[2] = "-p" argv[3] = "/dev/nst1" argv[4] = "beauty" argv[5] = "^/var/spool/mail$" argv[6] = "20010315" amrestore: 0: skipping start of tape: date 20010315 label WeeklySet112 amrestore: 1: skipping beauty._etc.20010315.0 amrestore: 2: restoring beauty._var_spool_mail.20010315.0 amrestore: read error: Input/output error amidxtaped: amrestore terminated normally with status: 2 Rewinding tape: done amidxtaped: pid 5970 finish time Fri Mar 16 11:33:44 2001 --- and the amindexd.debug file is --- amindexd: debug 1 pid 5966 ruid 422 euid 422 start time Fri Mar 16 11:31:04 2001 amindexd: version 2.4.2 220 redbaron AMANDA index server (2.4.2) ready. SECURITY USER root bsd security: remote host beauty.psc.sc.edu user root local user amanda amandahosts security check passed 200 Access OK DATE 2001-03-16 200 Working date set to 2001-03-16. SCNF WeeklySet1 200 Config set to WeeklySet1. HOST beauty.psc.sc.edu 501 No index records for host: beauty.psc.sc.edu. Invalid? HOST beauty.psc.sc.edu 501 No index records for host: beauty.psc.sc.edu. Invalid? HOST beauty 200 Dump host set to beauty. DISK / 501 No index records for disk: /. Invalid? DISK root_domain#root 501 No index records for disk: root_domain#root. Invalid? DISK /var/spool/mail - 2001-03-15 0 WeeklySet112 2 - 2001-02-23 0 WeeklySet110 2 - 2001-02-20 0 WeeklySet109 2 200 Disk set to /var/spool/mail. Uncompress command: /bin/gzip -dc '/usr/adm/amanda/index/beauty/_var_spool_mail/20010315_0.gz' 2/dev/null | sort '/usr/adm/amanda/index/beauty/_var_spool_mail/20010315_0' f /usr/adm/amanda/index/beauty/_var_spool_mail/20010315_0 200 "/" is a valid directory OISD / f /usr/adm/amanda/index/beauty/_var_spool_mail/20010315_0 200 "/" is a valid directory OLSD / 200- Opaque list of / 201- 2001-03-15 0 WeeklySet112 / 201- 2001-03-15 0 WeeklySet112 /adamt 201- 2001-03-15 0 WeeklySet112 /amanda 201- 2001-03-15 0 WeeklySet112 /copty 201- 2001-03-15 0 WeeklySet112 /kothari 201- 2001-03-15 0 WeeklySet112 /nussinov 201- 2001-03-15 0 WeeklySet112 /root 201- 2001-03-15 0 WeeklySet112 /singh 201- 2001-03-15 0 WeeklySet112 /sudheer 201- 2001-03-15 0 WeeklySet112 /yumiceva 200 Opaque list of / --- We have had a problem with ffs not working so we compiled with the broken ffs option but are still getting the same error message. Any ideas? Bill Campbell email: [EMAIL PROTECTED] homepage: http://hep.physics.sc.edu/~billc/
Re: exclude list format
I have a suggestion for you Andrew (of course it involves more work for you). A number of times I have seen on this list (and I have experienced it myself) that exclusion problems are related to broken versions of GNU tar. At a minimum, you should have a "Broken GNU tar" section, and let other people tell you what those broken versions are. [EMAIL PROTECTED] wrote: Any volunteers? (yeah, I thought not :-). I would be willing to write an EXCLUDE doc. It took me over two hours yesterday to figure this one out. The tar man page and a ps on the client were my friends here. I would need a least a week to allow my schedule to open up. Is there any start point yet? Andrew Hall -- Tom Schutter (mailto:[EMAIL PROTECTED]) Platte River Associates, Inc. (http://www.platte.com)
Re: exclude list format
Thankyou. I will add it. Could the list members please provide me details of this problem and the version of gnutar that have been known to cause this error. Thank you in advance. Andrew Hall On Fri, 16 Mar 2001, Tom Schutter wrote: I have a suggestion for you Andrew (of course it involves more work for you). A number of times I have seen on this list (and I have experienced it myself) that exclusion problems are related to broken versions of GNU tar. At a minimum, you should have a "Broken GNU tar" section, and let other people tell you what those broken versions are. [EMAIL PROTECTED] wrote: Any volunteers? (yeah, I thought not :-). I would be willing to write an EXCLUDE doc. It took me over two hours yesterday to figure this one out. The tar man page and a ps on the client were my friends here. I would need a least a week to allow my schedule to open up. Is there any start point yet? Andrew Hall -- Tom Schutter (mailto:[EMAIL PROTECTED]) Platte River Associates, Inc. (http://www.platte.com)
Re: A stupide question really
I have a really silly question for you. How can I check what is actually on the tape ? I mean is there a way to list the tape content ? It depends on what you want to know. If you just want a list of what images are on the tape (e.g. client and host), Gerhard's script will do that, or you can do this: mt -f /dev/whatever rewind amrestore -p /dev/whatever no-such-host /dev/null This searches the tape for client "no-such-host", which it will not find, but as it does so it will report everything it sees along the way. Note that Gerhard's script only works if your tape device is uses System V semantics. If it uses BSD, you'll need to do a little more work on it. If you want to know more about the tape, or to try to validate it a little more, you can use amverify. If you want to actually recover catalogues of all the files in all the images, there is no easy way to do that at the moment, although a script could probably be written to do so, probably based on amverify, which knows the proper dump program to run. Note that you often have to run the catalogue operation on the client as the host does not have the right program (for instance, a dump image of an AIX client cannot be cataloged on a Solaris tape server). Olivier Collet John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Re: exclude list format
I would be willing to write an EXCLUDE doc. ... Great! ... Is there any start point yet? I'd start with the GNU tar docs. A rough first cut would be to just take them and make a single document that could be released as part of Amanda. Then add more examples "typical" (whatever that means) of what Amanda users try to do (e.g. cut off a whole top level directory). Andrew Hall John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
file kills backup
hello amanda-users, i've run into a problem where my backup is failing because of a file it's finding that i can't seem to locate. i have reset amanda twice and it's failed both times. here is the failure notice from amanda: FAIL taper 192.168.1.11 /common/intranet 0 [input: can't open: 8^V@: No such file or directory] FAIL driver 192.168.1.11 /common/intranet 0 [dump to tape failed] i'm using gnutar on a redhat 6.2 box, with amanda 2.4.2, if that helps. amcheck finds no problems. the directory /common/intranet is accessible by netatalk and samba, so i'm thinking something got named strangely, but when i do a search for 8^V@, the search comes up empty. i've tried .8* and *8* (since i'm assuming the other characters are ascii and somehow not visable) but no luck. i'm very confused. can anyone tell me how to find and rename or delete the file, or how to get amanda to ignore it on the backup? thanks, walt
Failed backup of / partition
Hi! I am trying to setup amanda 2.4.2pl1 on two RedHat 6.2 boxes. The client has several HD partitions mounted, eg /,/usr,/home,/var etc. After I was successful in backing up one partition and restoring it, I scheduled a full dump of the client host. All of the partitions got backed up, except for the / partition: FAILURE AND STRANGE DUMP SUMMARY: brahma / lev 0 FAILED [disk / offline on brahma?] But / is NOT offline (otherwise the machine won't be running, would it?) Any suggestions? -Shreedeep = -+ Shreedeep [EMAIL PROTECTED] __ Do You Yahoo!? Get email at your own domain with Yahoo! Mail. http://personal.mail.yahoo.com/
Re: Failed backup of / partition
when this has happened to me, i find that a /etc/rc.d/init.d/inet restart command usually sets things correctly. run amcheck before and after to make sure it has been fixed. somehow the amandad stops accepting connections if there is a failure. reseting inet has worked for me. also, i have deleted the /tmp/amanda directory and then reset inet in extreme cases. hope that helps, w Shreedeep Bhachech wrote: Hi! I am trying to setup amanda 2.4.2pl1 on two RedHat 6.2 boxes. The client has several HD partitions mounted, eg /,/usr,/home,/var etc. After I was successful in backing up one partition and restoring it, I scheduled a full dump of the client host. All of the partitions got backed up, except for the / partition: FAILURE AND STRANGE DUMP SUMMARY: brahma / lev 0 FAILED [disk / offline on brahma?] But / is NOT offline (otherwise the machine won't be running, would it?) Any suggestions? -Shreedeep = -+ Shreedeep [EMAIL PROTECTED] __ Do You Yahoo!? Get email at your own domain with Yahoo! Mail. http://personal.mail.yahoo.com/
RE: amanda mail report size
On Fri, 16 Mar 2001, Ben Hyatt wrote:
So, in source /usr/local/src/amanda-2.4.1p1/example/config.site
add the CFLAGS entry, run configure and rebuild? (make clean, make make
install)
here's how i do it (for amanda, and similarly for other packages)...
mkdir /usr/local/src/amanda
mkdir /usr/local/src/amanda/2.4.2p1
create /usr/local/src/amanda/2.4.2p1/Makefile (attached)
create /usr/local/src/amanda/2.4.2p1/config.site (attached)
create /usr/local/src/amanda/2.4.2p1/amanda-2.4.2p1.tar.gz (distribution)
cd /usr/local/src/amanda/2.4.2p1
make ! 0-make-log
then you have a record of exactly how you configured and built amanda-X in
the Makefile and config.site. to rebuild at any time, just make again.
it's a very easy way to maintain local builds of software packages.
--
Todd Pfaff \ Email: [EMAIL PROTECTED]
Computing and Information Services \ Voice: (905) 525-9140 x22920
ABB 132 \ FAX: (905) 528-3773
McMaster University \
Hamilton, Ontario, Canada L8S 4M1 \
PACKAGE=amanda
VERSION=2.4.1p1
RELEASE=${PACKAGE}-${VERSION}
SHELL=/bin/sh
CONFIG_SITE=/usr/local/src/amanda/${VERSION}/config.site
#all: clean unpack patch build-client build-server install-client install-server
server: clean unpack patch build-server
client: clean unpack patch build-client
clean:
/bin/rm -rf ${RELEASE}
unpack:
umask 022;\
gnutar zxf ${RELEASE}.tar.gz;\
chown -R 0.0 ${RELEASE}
patch:
umask 022;\
patch -p0 -t patches/samba2-2418.diff
build-client:
umask 022;\
cd ${RELEASE};\
CONFIG_SITE=$(CONFIG_SITE); export CONFIG_SITE;\
./configure \
--without-server \
;\
make
build-server:
umask 022;\
cd ${RELEASE};\
CONFIG_SITE=$(CONFIG_SITE); export CONFIG_SITE;\
./configure \
;\
make
install-client:
umask 022;\
cd ${RELEASE};\
make install;\
cd /usr/local/amanda/lib;\
ln -s libamanda-2.4.1p1.so.0.0.0 libamanda-2.4.1p1.so;\
ln -s libamclient-2.4.1p1.so.0.0.0 libamclient-2.4.1p1.so;\
ln -s libamserver-2.4.1p1.so.0.0.0 libamserver-2.4.1p1.so;\
ln -s libamtape-2.4.1p1.so.0.0.0 libamtape-2.4.1p1.so;\
mkdir -p /usr/local/amanda/etc/gnutar-lists;\
chown backup.backup /usr/local/amanda/etc/gnutar-lists
install-server: install-client
install: install-client
# see srcdir/examples/config.site for details
CC=cc
CFLAGS=-DIGNORE_SMBCLIENT_ERRORS
sbindir=$prefix/bin
CONFIG_DIR=$prefix/etc
DEFAULT_SERVER=amandahost
FORCE_USERID=yes
CLIENT_LOGIN=backup
SETUID_GROUP=backup
USE_RUNDUMP=yes
SAMBA_CLIENT=/usr/local/samba/bin/smbclient
GNUTAR=$prefix/libexec/gnutar
GNUTAR_LISTDIR=$CONFIG_DIR/gnutar-lists
BSD_SECURITY=yes
USE_AMANDAHOSTS=yes
Re: file kills backup
i've run into a problem where my backup is failing because of a file it's finding that i can't seem to locate. i have reset amanda twice and it's failed both times. here is the failure notice from amanda: FAIL taper 192.168.1.11 /common/intranet 0 [input: can't open: À8^V@: No such file or directory] This message says taper was linking through the multiple holding disk chunks of the backup image of /common/intranet and one of them apparently has garbage in its header. Are you running amdump or amflush when this happens? If amflush, what's in your holding disk? Did anything odd happen when the dump happened into the holding disk, like file system full or anything like that? Can you just remove the holding disk chunks for this disk and force it to do a level 0 again on the next amdump run? walt John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Re: eof problem with amrecover
I keep getting the following error when I try to recover a file with amrecover: ... tar: Unexpected EOF on archive file ... amrestore: 0: skipping start of tape: date 20010315 label WeeklySet112 amrestore: 1: skipping beauty._etc.20010315.0 amrestore: 2: restoring beauty._var_spool_mail.20010315.0 amrestore: read error: Input/output error So what's your question? You got a tape read error. That's life. You can try all the usual things: * Clean the drive. * Retension the tape. * Try a different drive. * Check the cables, connectors, boards, etc. Bill Campbell John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Re: Failed backup of / partition
could it be that linux treats / partition (and therefor the disk that / belongs to) treats differently, as a result it doesn't let amanda "read" it? No. What did amcheck have to say? What's in /tmp/amanda/sendsize*debug on the client? Shreedeep John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Re: Missing backups?
Heres a transcript of the seession ... amrecover ls [?1h=[24;1H[K2001-03-15 . 2001-03-15 X11R6/ ... 2001-03-15 porWarning: `.' missing from directory ./ 2001-03-15 portWarning: `.' missing from directory ./ ... [24;1H[K[?1lamrecover cd home /usr/home amrecover ls [?1h=[24;1H[K2001-03-15 . [24;1H[K[?1lamrecover brown# ^Dexit Are you saying you got all this garbage instead of a normal listing of files? Sometimes it's hard to tell with E-mail whether what shows up is what was intended. Does anyone see any reason that I should not be getting backups? Why do you think you're not getting backups? Looks to me like backups worked fine but the index file is corrupt, which in turn makes amrecover's life miserable. First, I'd look through /tmp/amanda/amindexd*debug and see how it looks. If it shows the same corruption, then the problem is probably in the index files themselves, so I'd look at them next. If not, I'd look at /tmp/amnada/amrecover*debug and see what it thinks it got back from amindexd. In the meantime, if you really need the files restored soon, I'd "do it by hand" with amrestore. Stan Brown John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
re: file kills backup
This message says taper was linking through the multiple holding disk chunks of the backup image of /common/intranet and one of them apparently has garbage in its header. Are you running amdump or amflush when this happens? amdump, when i run amflush (only on days i haven't been here and no one changes the tape) if the directory hasn't been backed up i get: 192.168.1.11 -n/intranet NO FILE TO FLUSH - 192.168.1.11 /etc NO FILE TO FLUSH - 192.168.1.11 /var/namedNO FILE TO FLUSH - If amflush, what's in your holding disk? Did anything odd happen when the dump happened into the holding disk, like file system full or anything like that? Can you just remove the holding disk chunks for this disk and force it to do a level 0 again on the next amdump run? nothing odd happening, here's a df -k: Filesystem 1k-blocks Used Available Use% Mounted on /dev/hda8 256667 31465211950 13% / /dev/hda123302 3250 18849 15% /boot /dev/hda6 4245640 1209460 2820512 30% /home /dev/hda5 4245640389700 3640272 10% /usr /dev/hda7 256667 27428215987 11% /var /dev/hdd 14785660 2534400 11500176 18% /common my disklist for this machine is: 192.168.1.11 /common/intranet 192.168.1.11 /var/named 192.168.1.11 /etc and my holding disk is /common/amanda could it be confused because i'm using the same partition (but a separate directory) to back up? currently there is nothing in the holding disk, i flushed it this morning got the above message. this is the local machine that the amanda server resides on. it is able to back up the clients over the ethernet fine... i have 2 tapes run level 0 every night. it fails on either tape with the same error. thanks, w walt John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
Re: Failed dump
On Wed, 14 Mar 2001, John R. Jackson wrote: | I could start the backup while the things are calm within the same |procedure (same tape). It's a mail FS. As I told it's not constant. | |I still don't understand. Why would this file system be more idle at the |start of a backup? Note that it has nothing to do with other backups |being done. It's related to other file system activity unrelated to |backups (e.g. incoming mail, people reading mail, etc), most of which |you have no control over. My backup begins at 5:30am, becose I need to wait some other things to finish before start the backup. As I could see, the mail fs is the last to be done, at 9am when my system begin to be loaded. |If you know it's going to be more idle at a particular time, you can |use the "starttime" dumptype parameter to defer the backup to after a |particular time of day. It will be better if I could use it to anticipate not postpone the process. Thank you any way. - Marcelo
Re: Failed dump
My backup begins at 5:30am, becose I need to wait some other
things to finish before start the backup.
As I could see, the mail fs is the last to be done, at 9am when my
system begin to be loaded.
Ah, now that would be a problem. :-)
It will be better if I could use [starttime] to anticipate not postpone the
process.
The usual way to do this is to start the backup earlier but delay the
clients/disks that must "wait [for] some other things to finish".
For instance, let's say you are normally using dumptype root-tar.
Create a new dumptype like this:
defined dumptype root-tar-later {
root-tar# inherit all of root-tar
starttime 530 # wait until at least 5:30AM
}
Some of the file systems surely do not depend on the other things
finishing, including /var/mail, so let them use root-tar. They will start
(and finish) as soon as possible, hopefully before 5:30.
But change the other file systems that must be delayed to root-tar-later.
- Marcelo
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
RE: amanda mail report size
create /usr/local/src/amanda/2.4.2p1/Makefile (attached) create /usr/local/src/amanda/2.4.2p1/config.site (attached) create /usr/local/src/amanda/2.4.2p1/amanda-2.4.2p1.tar.gz (distribution) cd /usr/local/src/amanda/2.4.2p1 make ! 0-make-log Thank you very much Todd. then you have a record of exactly how you configured and built amanda-X in the Makefile and config.site. to rebuild at any time, just make again. it's a very easy way to maintain local builds of software packages. Makes a lot of sense I will have to start doing this for other software packages regards, -Ben Todd Pfaff \ Email: [EMAIL PROTECTED]
access error when using samba
I'm trying to backup a Win2000 machine from a Debian Linux 2.2 Amanda 2.4.2.p1/Samba 2.0.7 server. I get the following from amcheck: Amanda Backup Client Hosts Check ERROR: vedder: [PC SHARE //ozzy/ozzybkup access error: host down or invalid password?] Client check: 1 host checked in 20.467 seconds, 1 problem found -- Win2000 host name = ozzy samba share name = ozzybkup samba server host name amanda server/client = vedder user name (exists on ozzy and in smbpasswd) = backup -- AMANDA CONFIGURATIONS: Configured vedder with --with-smbclient=/usr/bin/smbclient. disklist contains: vedder.noemix.com //ozzy/ozzybkup windows-test /etc/amandapass contains: //ozzy/ozzybkup backup%same_password_as_in_smbpasswd -- SAMBA CONFIGURATIONS: Samba seems to be running okay: smbstatus Samba version 2.0.7 Service uid gid pid machine -- ozzybkup backup backup 22270 ozzy (10.1.1.52) Fri Mar 16 19:15:28 2001 No locked files Share mode memory usage (bytes): 1048464(99%) free + 56(0%) used + 56(0%) overhead = 1048576(100%) total -- Contents of smb.conf: [ozzybkup] comment = used by amanda to backup ozzy writable = yes locking = no path = /usr/local/share/ozzy public = yes Thanks, Terry
Re: access error when using samba
I'm trying to backup a Win2000 machine from a Debian Linux 2.2 Amanda 2.4.2.p1/Samba 2.0.7 server. I get the following from amcheck: Amanda Backup Client Hosts Check ERROR: vedder: [PC SHARE //ozzy/ozzybkup access error: host down or invalid password?] Go to vedder and look in /tmp/amanda/selfcheck*debug. It lists the smbclient command issued. Try running that by hand as the Amanda user. Terry John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
