Re: 2.6.1 client with 2.4.3 server?
Yes, this patch solved the problem. Thank you! Can we anticipate it being integrated into future versions? -Mitch On Tue, 19 May 2009, Jean-Louis Martineau wrote: amanda client 2.6.1p1 is compatible with server 2.4.3b4 and up. But it is not with 2.4.3b3 and previous. Try the attached patch. Jean-Louis Mitch Collinsworth wrote: Hi, Is there any gross imcompatability that would prevent a 2.6.1p1 client from being backed up by a 2.4.3b3 server? I've got it to the point where they are at least talking with each other, using auth bsd. amcheck -c is happy. estimates worked ok. But amstatus is showing this: client-a:/a0 driver: [parse of reply message failed] (10:34:30) client-a:/var 0 driver: [parse of reply message failed] (10:41:15) client-a:/var/log 0 31020k wait for dumping driver: (aborted:[request timeout]) hmm... here's something interesting in the sendbackup.debug file: 1242744076.458012: sendbackup: pid 14788 ruid 501 euid 501 version 2.6.1p1: start a t Tue May 19 10:41:16 2009 1242744076.458089: sendbackup: Version 2.6.1p1 1242744076.458650: sendbackup: sendbackup req: GNUTAR /var 0 1970:1:1:0:0:0 OPTI ONS |;bsd-auth;index; 1242744076.458736: sendbackup: Parsed request as: program `GNUTAR' 1242744076.458749: sendbackup: disk `/var' 1242744076.458760: sendbackup: device `/var' 1242744076.458770: sendbackup: level 0 1242744076.458781: sendbackup: since 1970:1:1:0:0:0 1242744076.458791: sendbackup: options `|;bsd-auth;index;' 1242744076.458998: sendbackup: start: client-a:/var lev 0 1242744076.459129: sendbackup: doing level 0 dump as listed-incremental to '/var/li b/amanda/gnutar-lists/client-a_var_0.new' 1242744076.461334: sendbackup: Started index creator: /bin/tar -tf - 2/dev/null | sed -e 's/^\.//' 1242744076.461537: sendbackup: pipespawnv: stdoutfd is 50 1242744076.461947: sendbackup: Spawning /usr/libexec/amanda/runtar runtar NOCONFIG /bin/tar --create --file - --directory /var --one-file-system --listed-incremental /var/lib/amanda/gnutar-lists/client-a_var_0.new --sparse --ignore-failed-read --to tals . in pipeline 1242744076.463151: sendbackup: gnutar: /usr/libexec/amanda/runtar: pid 14792 1242744076.463205: sendbackup: Started backup 1242744166.493587: sendbackup: critical (fatal): index tee cannot write [Broken pip e] /usr/lib64/amanda/libamanda-2.6.1p1.so[0x35834215b2] /lib64/libglib-2.0.so.0(g_logv+0x26f)[0x3af5c34d5f] /lib64/libglib-2.0.so.0(g_log+0x83)[0x3af5c34f33] /usr/libexec/amanda/sendbackup(start_index+0x290)[0x4037f0] /usr/libexec/amanda/sendbackup[0x4075f0] /usr/libexec/amanda/sendbackup(main+0x10a4)[0x405834] /lib64/libc.so.6(__libc_start_main+0xf4)[0x3fe1c1d8b4] /usr/libexec/amanda/sendbackup[0x403139] So it looks like a problem sending the index data. I've set index_server to the correct hostname in /etc/amanda-client.conf. Will this properly override the settings configured at compile time? This client is installed from the redhat rpm, so it has no local compile-time configuration. The /etc/amanda-client.conf looks like this: # # amanda.conf - sample Amanda client configuration file. # # This file normally goes in /etc/amanda/amanda-client.conf. # conf test # your config name index_server amanda-server# your amindexd server tape_server amanda-server# your amidxtaped server tapedev tape:/dev/nst1 # your tape device # if not set, Use configure or ask server. # if set to empty string , ask server # amrecover will use the changer if set to the value # of 'amrecover_changer' in the server amanda.conf. # auth- authentication scheme to use between server and client. # Valid values are bsd, bsdudp, bsdtcp, krb5, local, # rsh and ssh. # Default: [auth bsdtcp] auth bsd ssh_keys # your ssh keys file if you use ssh auth Anyone see anything obvious we're overlooking? (Yeah, besides needing to upgrade the server!) :-) -Mitch
2.6.1 client with 2.4.3 server?
Hi, Is there any gross imcompatability that would prevent a 2.6.1p1 client from being backed up by a 2.4.3b3 server? I've got it to the point where they are at least talking with each other, using auth bsd. amcheck -c is happy. estimates worked ok. But amstatus is showing this: client-a:/a0 driver: [parse of reply message failed] (10:34:30) client-a:/var 0 driver: [parse of reply message failed] (10:41:15) client-a:/var/log 0 31020k wait for dumping driver: (aborted:[request timeout]) hmm... here's something interesting in the sendbackup.debug file: 1242744076.458012: sendbackup: pid 14788 ruid 501 euid 501 version 2.6.1p1: start a t Tue May 19 10:41:16 2009 1242744076.458089: sendbackup: Version 2.6.1p1 1242744076.458650: sendbackup: sendbackup req: GNUTAR /var 0 1970:1:1:0:0:0 OPTI ONS |;bsd-auth;index; 1242744076.458736: sendbackup: Parsed request as: program `GNUTAR' 1242744076.458749: sendbackup: disk `/var' 1242744076.458760: sendbackup: device `/var' 1242744076.458770: sendbackup: level 0 1242744076.458781: sendbackup: since 1970:1:1:0:0:0 1242744076.458791: sendbackup: options `|;bsd-auth;index;' 1242744076.458998: sendbackup: start: client-a:/var lev 0 1242744076.459129: sendbackup: doing level 0 dump as listed-incremental to '/var/li b/amanda/gnutar-lists/client-a_var_0.new' 1242744076.461334: sendbackup: Started index creator: /bin/tar -tf - 2/dev/null | sed -e 's/^\.//' 1242744076.461537: sendbackup: pipespawnv: stdoutfd is 50 1242744076.461947: sendbackup: Spawning /usr/libexec/amanda/runtar runtar NOCONFIG /bin/tar --create --file - --directory /var --one-file-system --listed-incremental /var/lib/amanda/gnutar-lists/client-a_var_0.new --sparse --ignore-failed-read --to tals . in pipeline 1242744076.463151: sendbackup: gnutar: /usr/libexec/amanda/runtar: pid 14792 1242744076.463205: sendbackup: Started backup 1242744166.493587: sendbackup: critical (fatal): index tee cannot write [Broken pip e] /usr/lib64/amanda/libamanda-2.6.1p1.so[0x35834215b2] /lib64/libglib-2.0.so.0(g_logv+0x26f)[0x3af5c34d5f] /lib64/libglib-2.0.so.0(g_log+0x83)[0x3af5c34f33] /usr/libexec/amanda/sendbackup(start_index+0x290)[0x4037f0] /usr/libexec/amanda/sendbackup[0x4075f0] /usr/libexec/amanda/sendbackup(main+0x10a4)[0x405834] /lib64/libc.so.6(__libc_start_main+0xf4)[0x3fe1c1d8b4] /usr/libexec/amanda/sendbackup[0x403139] So it looks like a problem sending the index data. I've set index_server to the correct hostname in /etc/amanda-client.conf. Will this properly override the settings configured at compile time? This client is installed from the redhat rpm, so it has no local compile-time configuration. The /etc/amanda-client.conf looks like this: # # amanda.conf - sample Amanda client configuration file. # # This file normally goes in /etc/amanda/amanda-client.conf. # conf test # your config name index_server amanda-server# your amindexd server tape_server amanda-server# your amidxtaped server tapedev tape:/dev/nst1 # your tape device # if not set, Use configure or ask server. # if set to empty string , ask server # amrecover will use the changer if set to the value # of 'amrecover_changer' in the server amanda.conf. # auth- authentication scheme to use between server and client. # Valid values are bsd, bsdudp, bsdtcp, krb5, local, # rsh and ssh. # Default: [auth bsdtcp] auth bsd ssh_keys # your ssh keys file if you use ssh auth Anyone see anything obvious we're overlooking? (Yeah, besides needing to upgrade the server!) :-) -Mitch
Re: 2.6.1 client with 2.4.3 server?
On Tue, 19 May 2009, Jean-Louis Martineau wrote: I try to keep them compatible, but I don't test with release before 2.4.5. index_server is used only by amrecover. Can you post the amandad.*.debug file from the client? Jean-Louis Yup, here's a sample below. -Mitch 1242745188.258722: amandad: pid 14871 ruid 501 euid 501 version 2.6.1p1: start at Tue May 19 10:59:48 2009 1242745188.259342: amandad: security_getdriver(name=bsd) returns 0x358364e920 1242745188.259373: amandad: version 2.6.1p1 1242745188.259383: amandad: build: VERSION=Amanda-2.6.1p1 1242745188.259393: amandad:BUILT_DATE=Fri Apr 10 16:12:07 PDT 2009 1242745188.259402: amandad:BUILT_MACH=x86_64-unknown-linux-gnu BUILT_REV=1860 1242745188.259410: amandad:BUILT_BRANCH=amanda-261 CC=gcc 1242745188.259419: amandad: paths: bindir=/usr/bin sbindir=/usr/sbin 1242745188.259428: amandad:libexecdir=/usr/libexec 1242745188.259436: amandad:amlibexecdir=/usr/libexec/amanda mandir=/usr/share/man 1242745188.259445: amandad:AMANDA_TMPDIR=/tmp/amanda 1242745188.259453: amandad:AMANDA_DBGDIR=/var/log/amanda CONFIG_DIR=/etc/amanda 1242745188.259462: amandad:DEV_PREFIX=/dev/ RDEV_PREFIX=/dev/r 1242745188.259470: amandad:DUMP=/sbin/dump RESTORE=/sbin/restore VDUMP=UNDEF 1242745188.259479: amandad:VRESTORE=UNDEF XFSDUMP=UNDEF XFSRESTORE=UNDEF VXDUMP=UNDEF 1242745188.259489: amandad:VXRESTORE=UNDEF SAMBA_CLIENT=/usr/bin/smbclient 1242745188.259497: amandad:GNUTAR=/bin/tar COMPRESS_PATH=/usr/bin/gzip 1242745188.259506: amandad:UNCOMPRESS_PATH=/usr/bin/gzip LPRCMD=/usr/bin/lpr 1242745188.259515: amandad: MAILER=UNDEF listed_incr_dir=/var/lib/amanda/gnutar-lists 1242745188.259523: amandad: defs: DEFAULT_SERVER=localhost DEFAULT_CONFIG=DailySet1 1242745188.259532: amandad:DEFAULT_TAPE_SERVER=localhost DEFAULT_TAPE_DEVICE= 1242745188.259540: amandad:HAVE_MMAP NEED_STRSTR HAVE_SYSVSHM AMFLOCK_POSIX AMFLOCK_FLOCK 1242745188.259549: amandad:AMFLOCK_LOCKF AMFLOCK_LNLOCK SETPGRP_VOID ASSERTIONS 1242745188.259557: amandad:AMANDA_DEBUG_DAYS=4 BSD_SECURITY USE_AMANDAHOSTS 1242745188.259566: amandad:CLIENT_LOGIN=amandabackup CHECK_USERID HAVE_GZIP 1242745188.259574: amandad:COMPRESS_SUFFIX=.gz COMPRESS_FAST_OPT=--fast 1242745188.259582: amandad:COMPRESS_BEST_OPT=--best UNCOMPRESS_OPT=-dc 1242745188.259683: amandad: dgram_recv(dgram=0x3583659788, timeout=0, fromaddr=0x3583669780) 1242745188.259718: amandad: (sockaddr_in *)0x3583669780 = { 2, 769, 11.22.33.44 } 1242745188.259753: amandad: security_handleinit(handle=0xbcf5a30, driver=0x358364e920 (BSD)) 1242745188.262340: amandad: accept recv REQ pkt: SERVICE sendbackup OPTIONS hostname=client-a; GNUTAR / 0 1970:1:1:0:0:0 OPTIONS |;bsd-auth;index; 1242745188.262390: amandad: creating new service: sendbackup OPTIONS hostname=client-a; GNUTAR / 0 1970:1:1:0:0:0 OPTIONS |;bsd-auth;index; 1242745188.264233: amandad: sending ACK pkt: 1242745188.264340: amandad: dgram_send_addr(addr=0xbcf5a70, dgram=0x3583659788) 1242745188.264360: amandad: (sockaddr_in *)0xbcf5a70 = { 2, 769, 11.22.33.44 } 1242745188.264374: amandad: dgram_send_addr: 0x3583659788-socket = 0 1242745188.270256: amandad: security_streaminit(stream=0xbd06c60, driver=0x358364e920 (BSD)) 1242745188.270320: amandad: stream_server opening socket with family 2 (requested family was 2) 1242745188.270362: amandad: try_socksize: send buffer size is 65536 1242745188.270378: amandad: try_socksize: receive buffer size is 65536 1242745188.277672: amandad: bind_portrange2: Try port 11039: Available - Success 1242745188.277744: amandad: stream_server: waiting for connection: 0.0.0.0.11039 1242745188.277769: amandad: security_streaminit(stream=0xbd0f140, driver=0x358364e920 (BSD)) 1242745188.277789: amandad: stream_server opening socket with family 2 (requested family was 2) 1242745188.277817: amandad: try_socksize: send buffer size is 65536 1242745188.277831: amandad: try_socksize: receive buffer size is 65536 1242745188.285066: amandad: bind_portrange2: Try port 11039: Available - Address already in use 1242745188.292295: amandad: bind_portrange2: Try port 11040: Available - Success 1242745188.292368: amandad: stream_server: waiting for connection: 0.0.0.0.11040 1242745188.292392: amandad: security_streaminit(stream=0xbd171c0, driver=0x358364e920 (BSD)) 1242745188.292412: amandad: stream_server opening socket with family 2 (requested family was 2) 1242745188.292438: amandad: try_socksize: send buffer size is 65536 1242745188.292452: amandad: try_socksize: receive buffer size is 65536 1242745188.299718: amandad: bind_portrange2: Try port 11039: Available - Address already in use 1242745188.302654: amandad: bind_portrange2: Try port 11040: Available - Address already in
RE: Red Hat amanda question
On Wed, 4 Mar 2009, McGraw, Robert P wrote: For the OS file systems we use the native backup, which in this case is dump. Robert [Please remember to reply on-list when asking for help.] I was afraid of that. dump is not recommended or even expected to work on Linux. You need to use the tar dumptype and specify a file path rather than a device. The tar will run suid as root and the permissions problems won't occur. -Mitch -Original Message- From: owner-amanda-us...@amanda.org [mailto:owner-amanda- us...@amanda.org] On Behalf Of Mitch Collinsworth Sent: Tuesday, March 03, 2009 4:54 PM To: McGraw, Robert P Cc: amanda-users@amanda.org Subject: Re: Red Hat amanda question On Tue, 3 Mar 2009, McGraw, Robert P wrote: I have set up an Amanda client on a Red Hat 5.2 server and amcheck recognizes the new client. One of the things that I had to do for the 5.2 client to work is change the permissions and group for the device /dev/root. Are you doing dump or tar backups? -Mitch
Re: Red Hat amanda question
On Tue, 3 Mar 2009, McGraw, Robert P wrote: I have set up an Amanda client on a Red Hat 5.2 server and amcheck recognizes the new client. One of the things that I had to do for the 5.2 client to work is change the permissions and group for the device /dev/root. Are you doing dump or tar backups? -Mitch
Re: krb5 auth problem
If your realm is YZ.EDU, then that's what you use. If UVWX.YZ.EDU is a host name and not a realm name, then it doesn't belong in your principal names. Can you explain why you want to auth against the secondary rather than the primary? I can't think of any reason that should matter. -Mitch On Tue, 1 Jul 2008, Chad Kotil wrote: Heres an update to the kerberos realm issue I am now seeing. I want to use my secondary KDC (UVWX.YZ.EDU) rather than the primary KDC (YZ.EDU), but amanda doesnt seem to know how to look for it. I include the KDC realm in all of my config's. amanda.conf, and .k5login. Here is my .k5login backup/[EMAIL PROTECTED] I am able to kinit with the secondary KDC on the client using the keytab that I have on the server. [EMAIL PROTECTED] tmp]$ kinit backup/[EMAIL PROTECTED] -kt /home/ ckotil/keytab-amanda [EMAIL PROTECTED] tmp]$ This works just fine. Here is what i have in my amanda.conf krb5keytab /etc/amanda/keytab-amanda krb5principal backup/[EMAIL PROTECTED] The reason I think that amanda is ignoring the kerberos realm is because of this error that I see on the client in /tmp/amanda/amandad. 1214918169.546254: amandad: gss_name host/[EMAIL PROTECTED] 1214918169.546587: amandad: critical (fatal): gss_server failed: can't acquire creds for host key host/skip: No such file or directory It claims the gss_name is host/[EMAIL PROTECTED] when it should be host/[EMAIL PROTECTED] Any ideas? Thanks, --Chad On Jun 26, 2008, at 9:37 PM, Chad Kotil wrote: Ian, Jean-Loiuis provided me with a patch that fixed this problem. The patch was posted to the list. I now face a new problem. I need to use my secondary kdc REALM to authenticate, and not my default realm. The keytab on the server is from the second kdc realm and the principal is from this realm too. But, the client tries to authenticate with the default realm. Any idea how I can tell the client to use the secondary kerberos realm? Thanks, --Chad On Jun 26, 2008, at 6:46 PM, Ian Turner wrote: Chad, This is a bug in Amanda. I have filed a bug report. As a workaround, you can probably make it work by compiling --without-force-uid. I don't think we have any kerberos users who test out daily builds, so sometimes things break and nobody notices right away. Maybe if you have a spare machine, you can become the community kerberos tester. :-) Cheers, --Ian On Thursday 26 June 2008 10:36:56 you wrote: When i run spawn amandad via xinetd as root, i get this error. 1214490832.259079: amandad: critical (fatal): running as user root instead of amandabackup In the kerberos wiki it says amandad will relinquish root permissions after reading the keytab. It doesnt seem to be doing that. Also, What keytab on the client needs to be read as root? --Chad On Jun 25, 2008, at 5:29 PM, Jean-Louis Martineau wrote: xinetd must be configured to run amandad as root. Jean-Louis Chad Kotil wrote: I am trying to setup krb5 auth on amanda 2.6.0p1. I built the server and client --with-krb5-security, added a new principal to my KDC ([EMAIL PROTECTED] REALM), and wrote a keytab file and placed it on the server. It is locked down so only amandabackup (the user that runs amanda) can read it. The clients have a .k5amandahosts file containing the following: [EMAIL PROTECTED] REALM backupmaster.f.q.d.n [EMAIL PROTECTED] REALM my amanda.conf file contains krb5keytab /etc/amanda/krb5.keytab-amanda krb5principal [EMAIL PROTECTED] REALM On both of my krb5 auth clients I am seeing this error: 1214425629.641678: amandad: critical (fatal): gss_server failed: real uid is 10036, needs to be 0 to read krb5 host key 10036 is the UID for amandabackup, 0 is the UID for root. Both clients work fine if I just use bsdtcp auth. I am using ssh auth everywhere else but for these two particular hosts I cannot use ssh keys. Any ideas? Thanks, --Chad
Re: Can't seem to disable software compression.
compress none turns off compression of the data, but amanda will still compress the indexes. You can check your tape contents to verify that the data there is uncompressed. -Mitch On Wed, 30 Apr 2008, [UTF-8] Edi ??uc wrote: Hi. It seems that I can't disable software compression in amanda. I tried everything that could in my opinion have an impact on compression. But no luck. The problem is that i have a relatively slow machine and it takes forever to complete the backup. If I check the processes i see this: 19021 ?S 0:00 \_ /bin/sh /usr/sbin/amdump AAA1 19031 ?S 0:00 \_ /usr/lib/amanda/driver AAA1 19032 ?S 0:01 \_ taper AAA1 19037 ?S 0:01 | \_ taper AAA1 19033 ?S 0:08 \_ dumper0 AAA1 19034 ?S 5:31 \_ dumper1 AAA1 19074 ?S 0:08 | \_ /bin/gzip --best 19035 ?S 0:00 \_ dumper2 AAA1 19036 ?S 0:00 \_ dumper3 AAA1 19066 ?S 8:59 \_ chunker1 AAA1 I suppose that there shouldn't be a gzip --best if I disable the compression. I also tried to use the directive compress fast in the define dumptype section, but no luck. gzip --best is still there (it should at least change to gzip --fast). This is my disklist: server /boot tar_midp server / tar_midp And this is my amanda.conf: org AAA mailto root dumpuser backup # the user to run dumps under inparallel 4 dumporder sssS taperalgo first displayunit k netusage 600 Kbps dumpcycle 1 weeks runspercycle 5 tapecycle 10 tapes bumpsize 20 Mb bumppercent 20 bumpdays 1 bumpmult 4 etimeout 300 dtimeout 1800 ctimeout 30 tapebufs 20 usetimestamps no runtapes 1 tapedev /dev/nst0 rawtapedev changerfile /etc/amanda/AAA/changer.conf changerdev /dev/null maxdumpsize -1 tapetype HPC7972A labelstr ^AAADLT[0-9][0-9]*$ amrecover_do_fsf yes amrecover_check_label yes amrecover_changer holdingdisk hd1 { directory /var/local/arhiv/amanda_hold use -2 Gb chunksize 2Gb } autoflush no infofile /var/lib/amanda/AAA1/curinfo logdir /var/log/amanda/AAA1 indexdir /var/lib/amanda/AAA1/index define tapetype HPC7972A { comment just produced by tapetype program length 193024 mbytes filemark 0 kbytes speed 13723 kps } define dumptype global { comment Global definitions index yes compress none holdingdisk auto comprate 1 estimate calcsize } define dumptype tar_midp { comment Tar mid priority program GNUTAR exclude list .amanda.excludes priority medium compress none index yes holdingdisk auto comprate 1 estimate calcsize } define interface local { comment a local disk use 1000 kbps } define interface le0 { comment 10 Mbps ethernet use 400 kbps } The installation is on Debian etch (4.0). Amanda (server and client) version 2.5.1p1-2.1 Anyone any idea? Regards, Edi
Re: backup/recover using tar and hard links
On Wed, 17 Oct 2007, Dustin J. Mitchell wrote: Obviously, this isn't ideal. I'm surprised nobody else has been snagged by this before. I've been using Amanda on Cyrus mailboxes for years, with lots of recoveries. I guess I've just gotten lucky. We have no experience with cyrus, yet, but have been talking about it for a while. After reading this I forwarded it to a co-worker, who wrote back this: Don't let cyrus do it. From http://cyrusimap.web.cmu.edu/imapd/overview.html#singleinstance -- Single Instance Store If a delivery attempt mentions several recipients (only possible if the MTA is speaking LMTP to lmtpd), the server attempts to store as few copies of a message as possible. It will store one copy of the message per partition, and create hard links for all other recipients of the message. Single instance store can be turned off by using the singleinstancestore flag -- Obviously this won't help the person with the failing restore, but it's something to consider for everyone else. Who really is short enough on disk space in this day and age that they think storing duplicate e-mail with hard links is still a good idea? (And just btw, we had lately been losing the battle against spam using spamassassin, even though it was pretty effective when first deployed. We recently added graylisting and are seeing very few spam making it through now.) -Mitch
Re: Sun 622 DLT8000 drive
On Wed, 19 Sep 2007, Toomas Aas wrote: I'm thinking about buying an used DLT8000 drive for the puropse of occasionally reading some old DLT tapes (not with Amanda, so sorry for the OT). Sun Model Number 622, Part Number 599-2347-02 is available on eBay, but I can't find any online documentation about this specific drive. Maybe someone on this list has one, and can point me to some online docs? I assume that it is basically a standard SCSI DLT drive and should work fine with non-Sun server, right? Right. It should be able to read any tape written by DLT4000, DLT7000, or DLT8000. -Mitch
Re: ssh tunneling from wherever
On Tue, 7 Aug 2007, Dustin J. Mitchell wrote: All in all, it sounds like a lot of work, unless this is a months-long conference :) Which is why it would be really nice to have a different triggering method for performing backups on roaming laptops. Something that begins with the laptop calling in to the server and saying Yoo-hoo! I'm over here. Can you please back me up now? Then the server can do the backup to holding disk and then flush to tape next time a regular scheduled config run is made. -Mitch
Re: new backup server
On Thu, 14 Dec 2006, Frank Smith wrote: AIT5 recently came out, and it can read AIT3 and AIT4 tapes, and has 400GB native capacity. Interesting. I followed up on this and it appears to be true. So, we will consider upgrading our Qualstar AIT2 library with AIT5 drives at some point. Still, we're going forward with our LTO3 library purchase at this point. The one failure I see in reading the AIT5 specs is that is it 400 GB native capacity but still 24 MB/sec transfer rate, same as AIT4. So it will take nearly 5 hours to write a tape at full speed. After waiting forever for AIT3 to get out the door, and after then being promised that AIT4 would be backwards compatible, only to find out it wasn't sometime AFTER it was released, we've lost all loyalty to the AIT line. Unless you frequently have a need to read old tapes, keeping a an old drive or two around just to read old tapes isn't a big deal. I gotta disagree with this though. Keeping an old AIT drive around to be able to infrequently read old tapes is a recipe for disappointment. Old AIT drives' capstans eventually fail and lead to drive errors. If you want to read old AIT tapes, either use a modern drive with backwards read compatibility, plan on having to refurbish that old drive you're keeping around every now and then, or migrate the data to modern media. The advantage of not switching formats is that you can just replace the drives and the tapes to upgrade a library to higher capacity. Agreed. Which is the only reason we will consider buying AIT5 drives now, because we already have an AIT2 and an AIT3 library. AIT4 was, apparently, just a very bad dream. -Mitch
Re: new backup server
On Thu, 14 Dec 2006, Chris Hoogendyk wrote: I'm interested in whether anyone on the list has any experience or comments on my choice of tape changer, or comments on issues related to how it is configured and potential modes of upgrading (adding another tape drive, adding another changer, etc.) You didn't say which AIT drive is going in your AIT changer. Here we have gone from AIT1 to AIT2 to AIT3. Just yesterday I ordered a new library with LTO3. What soured us on the AIT line is that AIT4 is not backwards read compatible with any earlier AIT drives. In other words if I went to AIT4 I would not be able to use it to even read any of our large existing collection of AIT1, 2, and 3 tapes. So at this point it no longer matters to us whether we stay with the AIT line or not. Depending on which AIT drive you're choosing, this may or may not be a concern for you. Given that it is for us, we took this as our opportunity to move to LTO, which is at least an industry standard with multiple vendors supplying drives. (Sony can take as long as they want to come out with the next generation of AIT, since they're the only supplier. We waited what seemed like forever for AIT3 to finally come out. Way past its expected release date. And AIT4 was promised all along to be backwards read compatible, but that was dropped at the very last minute.) The tape changer I'm looking at is the Sony StorStation AIT Library LIB-162/A4. It is a carousel rather than a robot. It holds 16 tapes (3.2TB native, anybody's guess compressed) and can have a second tape drive added. It is significantly less expensive than the expandable robot systems I was looking at. Also, in the expandable systems, adding the expansions was very expensive. Not sure what systems you looked at, but I was surprised to find that in the Qualstar RLS series of expandable libraries, adding more tape slots is not a big money proposition. The LTO library I ordered starts with 12 slots and is expandable up to 44 slots in increments of 8, for $1000 (list) per increment. As an .edu you may do better than that on price. Also with AIT slots being smaller, they might come cheaper, too. I don't know. Hope this helps in some way. -Mitch
Re: dd problems
I am trying to use dd to copy a tape made by amanda to a temp spot on the HDD, but I am having great difficulty. When I try to use dd, it will only copy one file at a time then stop, no matter what I do. As far as I know, dd should just take the entire tape from beginning to end. . . dd does one tape file at a time, which yes, is what amanda writes. In order to do what you're trying to do you just need to write a short script to put your dd into a loop. -Mitch
Re: amdump issues involving wildcards--quoted wildcards appear to be sent to, and thus being interpreted literally by, gtar
On Thu, 2 Nov 2006, Ian R. Justman wrote: My conjecture of what's happening is that the gtar command is being run with quotes around the items in the include statement in my disklist file, causing gtar to not perform any globbing and, instead, look for any pathname with actual brackets and globbing characters in it. Have a look in /tmp/amanda/sendbackup.xxx.debug. That should show you what the command line arguments being used look like. -Mitch
Re: Still struggling with L0
On Mon, 31 Jul 2006, Alan Pearson wrote: I've tried to force a L0 backup of it, like : amadmin DailySet1 force qtvpdc.lab:/Shares amadmin: qtvpdc.lab:/Shares is set to a forced level 0 at next run. But when amdump runs, it just schedules a L1 or L2 backup. Syntax: % amadmin help Usage: amadmin conf command {args} ... Valid commands are: version # Show version info. force [hostname [disks]* ]+ # Force level 0 at next run. unforce [hostname [disks]* ]+ # Clear force command. force-bump [hostname [disks]* ]+# Force bump at next run. force-no-bump [hostname [disks]* ]+ # Force no-bump at next run. unforce-bump [hostname [disks]* ]+ # Clear bump command. reuse tapelabel ... # re-use this tape. no-reuse tapelabel ...# never re-use this tape. find [hostname [disks]* ]* # Show which tapes these dumps are on. delete [hostname [disks]* ]*# Delete from database. info [hostname [disks]* ]* # Show current info records. due [hostname [disks]* ]* # Show due date. balance # Show nightly dump size balance. tape# Show which tape is due next. bumpsize# Show current bump thresholds. export [hostname [disks]* ]*# Export curinfo database to stdout. import # Import curinfo database from stdin. disklist [hostname [disks]* ]* # Show disklist entries. So your amadmin DailySet1 force qtvpdc.lab:/Shares should be amadmin DailySet1 force qtvpdc.lab /Shares -Mitch
Re: Incorrect bahaviour causes backup loss - further update!!
On Fri, 28 Jul 2006, Alan Pearson wrote: How can I force a level 0 dump ? amadmin force
Re: Getting list of DLE's w/o Lev 0s
On Fri, 28 Jul 2006, C. Chan wrote: Is there a quick way to determine which DLEs do not have any Lev 0s on any tapes? amoverview
Re: Incorrect bahaviour causes backup loss !!
[Please don't cross-post.] It sounds like you don't have enough tapes in your rotation. You probably want to look into adding more. Also, amcheck will alert you to this problem before you amdump. If you choose to ignore the warning... -Mitch
Re: Incorrect bahaviour causes backup loss !!
[Please don't cross-post.] On Thu, 27 Jul 2006, Alan Pearson wrote: It happened because some large files were added in between amcheck and amdump, and over the past month more machines added, taking us nearer the tape capacity. I will add more tapes (when they arrive !!), but it _could_ still happen, and I believe it should NEVER be allowed to happen that you get into a state where you don't have a full backup of the machine. Well let's consider your options... If the next tape in sequence has a dump you don't want overwritten, then you need to pull it out of the sequence and put it in a drawer somewhere to keep it safe. Then you'll want to go to the next tape in the sequence. But if your tape cycle is too short you're likely to have the same problem with that one, too. If your plan is to write to holding disk until a new box of tapes arrive then don't put any tape in the drive. If you're this short on space (and yes I know it happens, it's happening to me right now, too) then you have to monitor things closely. Keep an eye on amcheck output, your daily amdump e-mails, and on amoverview. I really doubt you're going to find a majority of amanda users who agree with changing the system to write nothing to the tape that's in the drive when it's the correct next tape in sequence, just because your data doesn't fit on the number of tapes you've providing. -Mitch
Re: One DLE too many!
On Wed, 15 Mar 2006, Jean-Louis Martineau wrote: That'a not fixed in 2.5.0, that will be done after the dumper-api get implemented. dumper-api? I thought the story was that's being replaced with application-api. In any case I offered a programmer's time 2 years ago to fix this w/o waiting on any api. I just wanted agreement first that if we developed it as proposed it would be accepted. But we were never told it would be accepted. Or not accepted. Just no answer at all. -Mitch
Re: One DLE too many!
On Tue, 14 Mar 2006, Jon LaBadie wrote: On Tue, Mar 14, 2006 at 01:44:50PM -0500, Vytas Janusauskas wrote: Amanda Backup Client Hosts Check ERROR: hal: [Can't open disk '/mnt/data06/Deforest3'] ERROR: hal: [No include for '/mnt/data06/Deforest3_PART1'] Client check: 5 hosts checked in 3.259 seconds, 2 problems found IIRC amcheck does NOT run some of its checks as root. Thus if the amanda user running amcheck can not visit /mnt/data06/Deforest3 and needed directories below that, it could cause errors like that above when amcheck looked for files. It's worse than that. Include processing is performed before setuid root mode is started. selfcheck, sendsize, and sendbackup all fail to include directories that the amanda user can't read. You just lose and the directories don't get backed up. Makes it hard to provide a backup service for machines that aren't centrally managed, which is more and more the way the world (for some of us) is going. -Mitch
Re: One DLE too many!
On Tue, 14 Mar 2006, Stefan G. Weichinger wrote: Mitch Collinsworth schrieb: On Tue, 14 Mar 2006, Jon LaBadie wrote: IIRC amcheck does NOT run some of its checks as root. Thus if the amanda user running amcheck can not visit /mnt/data06/Deforest3 and needed directories below that, it could cause errors like that above when amcheck looked for files. It's worse than that. Include processing is performed before setuid root mode is started. selfcheck, sendsize, and sendbackup all fail to include directories that the amanda user can't read. You just lose and the directories don't get backed up. Makes it hard to provide a backup service for machines that aren't centrally managed, which is more and more the way the world (for some of us) is going. Sounds BAD. I'd like to know more on that ... is this a major issue for Amanda in general ... do we have to do major patches ... what can we do ... It's BAD for those of us who are trying to offer backups for machines that aren't centrally managed. It was discussed a bit on -hackers in 2004 - April and again in July. Solutions were suggested but never agreed upon. One was to add another setuid to perform include parsing. Another was to add include parsing to gnutar. Amanda 2.5.0 is near and I would really prefer to get it out without major problems in its source code. I haven't looked at 2.5.0 code to see if it's fixed there or not. -Mitch
Re: One DLE too many!
On Tue, 14 Mar 2006, Stefan G. Weichinger wrote: Mitch Collinsworth schrieb: It's BAD for those of us who are trying to offer backups for machines that aren't centrally managed. It was discussed a bit on -hackers in 2004 - April and again in July. Solutions were suggested but never agreed upon. One was to add another setuid to perform include parsing. Another was to add include parsing to gnutar. I don't remember that thread: - What exactly do you mean by (not) centrally managed? Sorry. What I mean is, in the case where the admin for the backup system is also the admin for the machines being backed up (or at least works in the same shop with that admin) it is reasonably easy to work around this problem by just modifying the directory permissions to allow the amanda user to read what it needs. Easier than trying to solve the problem by modifying amanda. In the case where the backup admin is offering a service to folks who admin their own machines, each time this problem is encountered it takes a bunch of time to walk the user through making the permission changes. And of course one of the questions one often hears in this situation is I have to fiddle my permissions in order for your backup software to work? Are you sure this is the best software for this? Modifying amanda to avoid the problem in the first place looks much more attractive in this situation. -Mitch
Re: Off-Topic: gmail problems with the list
On Thu, 16 Feb 2006, Guy Dallaire wrote: Could the list insert it's e-mail address in the reply-to header of it's messages ? Hack the list to work around a missing feature in your e-mail client? Please, no. Otherwise, when there's nothing, I thing gmail uses the Original poster's email adress for the reply to. Is there really no reply-all function? If not then you should find a better e-mail client. -Mitch
Re: Running multiple amdumps in paralell?
On Mon, 12 Sep 2005, Toralf Lund wrote: Is it safe to run multiple instances of amdump simultaneously? I mean, with different configs, but possibly the same hosts and disks? Different configs: yes same client hosts: no -Mitch
Re: Holding disk size question
On Sat, 27 Aug 2005, Marcus wrote: In amanda.conf it says If a dump is too big to fit on the holding disk than it will be written directly to tape. I'm hoping to write 200g at a time to DLT, and would like to stream it if possible. Does that mean I need 200g of holding disk? Yep. I scrounged up 20g, but another 180g doesn't look likely. Huh? Disks are dirt cheap these days. Look at this one: http://www.newegg.com/product/Product.asp?Item=N82E16822144309CMP=OTC-FroogleATT=Western+Digital+Caviar+SE+WD2500JB+250GB+7200+RPM+IDE+Ultra+ATA100+Hard+Drive 250 GB for $110. That's nothing compared to what you'll be spending on a rack of tapes, and what you probably already paid for the DLT drive. -Mitch
Re: [off-topip] Better Backup Media
On Tue, 21 Jun 2005, Chris Loken wrote: BUT - my AIT-4 drive apparently can't even read AIT-3 tapes. Seems mind-bogglingly stupid but my vendor assures me it's true (haven't dared to try). Anybody understand if this is an issue that's going to be resolved and, if so, will it be in firmware or hardware? Answer looks pretty clear from the charts here: http://b2b.sony.com/documents/category/storage/branded-tape/AIT_Drives/AIT-4/AITMedia_05.pdf I suppose it was necessary in order to progress, but bailing on backwards compatibility is very annoying to the customer. It's also not especially good for business. If I've been buying all the AIT line drives, at least in part because the newer ones will read tapes I wrote with the older ones, then I have a business reason to engage in brand loyalty. Once that reason is gone then the next time I upgrade drives, all media choices are back on the table again and I may choose something other than AIT the next time. Didn't the same thing happen with DLT between DLT8000 and SDLT? -Mitch
Re: [off-topip] Better Backup Media
On Thu, 23 Jun 2005, Joshua Baker-LePain wrote: Something else I heard on another mailing list: However AIT-4 (unlike AIT-1 til -3) appears to write fill bytes onto the tape if it's not fed with data quickly enough, thus wasting lots of capacity. The person said they heard it somewhere and hadn't seen it confirmed. If true, though... yuck. Well, maybe, maybe not. DLT8000 did that, too. Lots of people griped bitterly about poor performance with their DLT4000's and 7000's. Problem was usually that their 1-pass backup software kept starving the write buffer and the drives had to shoe-shine in order to deal with it. The DLT8000 had variable speed write, which meant the tape kept streaming and the data was laid down as fast as it came in, even if that was slower than what the tape could handle. I always had a good chuckle at conferences listening to vendors trying to explain this problem to all the folks griping about their expensive tape drives that would only write at a fraction of their advertised speed. In general the vendors would do everything they could to avoid pointing the blame at the expensive commercial backup software products because they usually sold that to the customers, too. The good news for amanda users is that when you stage your dumps to holding disk, you eliminate the most frequent cause of the data starvation problem, which is the backup program scouring the partition looking for which files to backup today. Once you have your data on the holding disk you're unlikely to starve the tape drive and it can stream the data at full speed onto the tape. If it can't, you have a h/w problem that is typically easy to fix. -Mitch
Re: Tape Library Recommendations
On Thu, 23 Jun 2005, Michael Loftis wrote: LTO, DLT, S-LTO, etc all have the huge advantage of the same physical form factor. So 'upgrading' a DLT library to LTO, or S-LTO is just adding/upgrading the tape drives in it. DLT (sometimes called Compactape IV) has a long history and is a good reliable medium, with a lot of vendors selling drives and Yeah that's what you'd think. I used to run an Overland loader with DLT4000. The thing was rock-solid, so when the time came I tried to swap out the drive for a DLT7000. Nope, Overland wouldn't hear of it. They insisted I had to start all over with a new loader. (Yeah, much more $$, too.) Turned out at least part of the problem was that the SCSI version of the 4000 loader wasn't fast enough for the 7000 drive. -Mitch
Re: [off-topip] Better Backup Media
On Thu, 23 Jun 2005, Joshua Baker-LePain wrote: Indeed -- the other person referred to the AIT-4 behavior as DLT syndrome. But isn't variable speed write different than writing fill bytes? Does DLT8000 lose capacity when not writing as fast as it can? Yes. Same idea. Tape spins at constant rate, data is written as fast as possible or else as fast as it comes in. They probably had some minimum below which it would stop and restart, but you'd get more on a tape if you could keep the write buffer from starving. Actually, with LTO3, I'm a bit worried about keeping the tape streaming when writing from staged dumps. Native transfer rate of LTO-3 is stated as 288 GB/hr, which is about 76 MiB/s. That's more than most single spindles can handle, *especially* if you're trying to write to tape while other dumps are coming in to the holding disk. Looks like I'll have to do some work on the server end to really get this thing going. Wow, that's impressive. First question is: what does the drive do when you can't feed it data fast enough? Depending on the answer you may or may not actually care enough to worry about it. -Mitch
Re: Amanda Statistics.
On Thu, 16 Jun 2005, Erik P. Olsen wrote: Below the statistics from my last backup. I assume the times are all elapsed times, but why is amanda so wrong in estimating the times? Or is it a bug? STATISTICS: Total Full Daily Estimate Time (hrs:min)0:03 Run Time (hrs:min) 1:11 Dump Time (hrs:min)0:51 0:51 0:01 Output Size (meg)7692.4 7648.9 43.5 Original Size (meg) 11996.711886.5 110.2 Avg Compressed Size (%)64.1 64.4 39.4 (level:#disks ...) Filesystems Dumped7 2 5 (1:5) Avg Dump Rate (k/s) 2559.5 2584.4 950.4 Tape Time (hrs:min)0:49 0:49 0:00 0:03 is not the estimated time for the run, it's how long it took to perform the estimate phase of the run. -Mitch
Re: Levels
On Thu, 12 May 2005, Guy Dallaire wrote: I'm still having a hard time figuring out how levels work. Can someone confirm to me that each time amanda is run, it will backup a file if it has been modified during the day (or since the last run) In other words: does amanda guarantee that if a file is modified after a dump, it will be included in the next dump, whatever the level is used ? No. Amanda guarantees it will take all data your backup program sends it, put it onto the tape, and get it back off again, assuming the tape is still good at that time. Whether or not a modified file gets included in the next dump is the responsibility of whatever backup program you tell amanda to invoke. Typically dump or gnutar, but folks have managed to glue in a number of others as well. Amanda doesn't do backups, she only oversees the process. -Mitch
Re: Levels
[Please reply to amanda-users when asking questions on the list] On Thu, 12 May 2005, Guy Dallaire wrote: 2005/5/12, Mitch Collinsworth [EMAIL PROTECTED]: No. Amanda guarantees it will take all data your backup program sends it, put it onto the tape, and get it back off again, assuming the tape is still good at that time. Whether or not a modified file gets included in the next dump is the responsibility of whatever backup program you tell amanda to invoke. Typically dump or gnutar, but folks have managed to glue in a number of others as well. Amanda doesn't do backups, she only oversees the process. -Mitch Well... What I meant is that, provided I have a standard dump cycle, I use gnutar for some DLE's, and dump for other DLE's, and I run amanda at time x. Will amanda run dump or gnu-tar in a way that if a file is modified before time x, it will ALWAYS be included in the backup ? ALWAYS is a very strong word. I see others have been responding yes to this but nevertheless the answer is still no. Amanda makes a very good effort to always do the right thing but there are various ways in which things can still go wrong. The best you can really say is that amanda will ALWAYS try. For example if you have way too much data to fit on the tape, sometimes amanda will say [dumps way too big, must skip incremental dumps]. Now this is generally a very rare occurance and typically one that is not caused by any failing in amanda, only that you gave her more to do than is physically possible and she had to decide SOMETHING. Also note that there are ways to configure your amanda setup to make this highly unlikely. For example a tape loader with plenty of tapes available and a configuration that tells amanda she can use multiple tapes per run if necessary. etc. -Mitch
Re: Amanda and Star
On Mon, 25 Apr 2005, Jon LaBadie wrote: The only way I know of to get an accurate backup of windows systems is to use a Windows-based backup program that outputs its results into a file. Then that file can be saved onto tape by amanda. Restores would similarly be an unpleasant, two-step process. This seems to be one place where Bacula has the advantage over Amanda. -Mitch
mail.vh.org is re-sending old mail to amanda-users list
Todd, mail.vh.org is re-sending old messages (from earlier today) to the amanda-users mailing list. Example below. It should probably be blocked from posting until this is fixed. vh.org postmaster, please fix your system. -Mitch -- Forwarded message -- Return-Path: [EMAIL PROTECTED] Received: from mercury.ccmr.cornell.edu [128.84.231.97] by twin.ccmr.cornell.edu with POP3 (fetchmail-6.2.1) for [EMAIL PROTECTED] (single-drop); Fri, 22 Apr 2005 17:01:01 -0400 (EDT) Received: from guinness.omniscient.com (guinness.omniscient.com [64.134.101.78]) by mercury.ccmr.cornell.edu (8.13.1/8.12.10) with ESMTP id j3MKweY8010571 for [EMAIL PROTECTED]; Fri, 22 Apr 2005 16:58:40 -0400 Received: from guinness.omniscient.com (localhost [127.0.0.1]) by guinness.omniscient.com (8.13.3/8.13.1) with ESMTP id j3MKwcgb020690 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Fri, 22 Apr 2005 16:58:38 -0400 (EDT) Received: from localhost ([EMAIL PROTECTED]) by guinness.omniscient.com (8.13.3/8.12.11/Submit) with SMTP id j3MKwcws000871; Fri, 22 Apr 2005 16:58:38 -0400 (EDT) Received: by guinness.omniscient.com (bulk_mailer v1.13); Fri, 22 Apr 2005 16:47:40 -0400 Received: from guinness.omniscient.com (localhost [127.0.0.1]) by guinness.omniscient.com (8.13.3/8.13.1) with ESMTP id j3MKTGfl024158 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Fri, 22 Apr 2005 16:29:17 -0400 (EDT) Received: (from [EMAIL PROTECTED]) by guinness.omniscient.com (8.13.3/8.12.11/Submit) id j3MKTFfB006444; Fri, 22 Apr 2005 16:29:15 -0400 (EDT) X-Authentication-Warning: guinness.omniscient.com: majordom set sender to [EMAIL PROTECTED] using -f Received: from sapphire.omniscient.com (sapphire.omniscient.com [64.134.101.71]) by guinness.omniscient.com (8.13.3/8.13.1) with ESMTP id j3MKTCE3005050 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for amanda-users@amanda.org; Fri, 22 Apr 2005 16:29:12 -0400 (EDT) Received: from mail.vh.org (mail.vh.org [129.255.233.40]) by sapphire.omniscient.com (8.13.1/8.13.1) with SMTP id j3MKTAhG015538 for amanda-users@amanda.org; Fri, 22 Apr 2005 16:29:10 -0400 (EDT) Received: (qmail 2652 invoked by uid 501); 22 Apr 2005 20:28:54 - MBOX-Line: From [EMAIL PROTECTED] Fri Apr 22 15:28:54 2005 Received: (qmail 22870 invoked by uid 18); 22 Apr 2005 19:48:55 - Received: from [EMAIL PROTECTED] by mail by uid 131 with qmail-scanner-1.20rc4 (clamuko: 0.80 Clear:RC:0:. Processed in 0.052428 secs); 22 Apr 2005 19:48:55 - Received: from guinness.omniscient.com (64.134.101.78) by mail.vh.org with SMTP; 22 Apr 2005 19:48:55 - Received: from guinness.omniscient.com (localhost [127.0.0.1]) by guinness.omniscient.com (8.13.3/8.13.1) with ESMTP id j3MJms5R026943 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Fri, 22 Apr 2005 15:48:54 -0400 (EDT) Received: from localhost ([EMAIL PROTECTED]) by guinness.omniscient.com (8.13.3/8.12.11/Submit) with SMTP id j3MJmph5013477; Fri, 22 Apr 2005 15:48:51 -0400 (EDT) Received: by guinness.omniscient.com (bulk_mailer v1.13); Fri, 22 Apr 2005 15:46:55 -0400 Received: from guinness.omniscient.com (localhost [127.0.0.1]) by guinness.omniscient.com (8.13.3/8.13.1) with ESMTP id j3MJktuc020306 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Fri, 22 Apr 2005 15:46:55 -0400 (EDT) Received: (from [EMAIL PROTECTED]) by guinness.omniscient.com (8.13.3/8.12.11/Submit) id j3MJktwA005561; Fri, 22 Apr 2005 15:46:55 -0400 (EDT) X-Authentication-Warning: guinness.omniscient.com: majordom set sender to [EMAIL PROTECTED] using -f Received: from sapphire.omniscient.com (sapphire.omniscient.com [64.134.101.71]) by guinness.omniscient.com (8.13.3/8.13.1) with ESMTP id j3MJkscn021387 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for amanda-users@amanda.org; Fri, 22 Apr 2005 15:46:54 -0400 (EDT) Received: from loncoche.terra.com.br (loncoche.terra.com.br [200.154.55.229]) by sapphire.omniscient.com (8.13.1/8.13.1) with ESMTP id j3MJkqeK016062 for amanda-users@amanda.org; Fri, 22 Apr 2005 15:46:53 -0400 (EDT) Received: from estero.terra.com.br (estero.terra.com.br [200.154.55.138]) by loncoche.terra.com.br (Postfix) with ESMTP id 6DA4EE78AF6 for amanda-users@amanda.org; Fri, 22 Apr 2005 16:46:46 -0300 (BRT) X-Terra-Karma: 0% X-Terra-Hash: 3f16c97bdda1a80776a43a2705e3fcab Received-SPF: pass (estero.terra.com.br: domain of terra.com.br designates 200.154.55.138 as permitted sender) client-ip=200.154.55.138; [EMAIL PROTECTED]; helo=terra.com.br; Received: from terra.com.br (tupiza.terra.com.br [200.176.3.182]) (authenticated user vlpg) by estero.terra.com.br (Postfix) with ESMTP id 59282728025 for amanda-users@amanda.org; Fri, 22 Apr 2005 16:46:45 -0300 (BRT) Date: Fri, 22 Apr 2005
RE: Amanda vs Homegrown
On Thu, 21 Apr 2005, Mark Lidstone wrote: It would still be worth pointing out what a huge security risk the rcp command is, and if they insist on using their scripts at least get them to remove the r* accounts setup stuff and use something like rsync over an encrypted channel (why bother protecting the file on the disk if you're going to potentially transfer it in plain text over the network). So have you modified amanda to encrypt your network transfers? It doesn't do that out of the box you know. -Mitch
Re: Changer file for Compaq MSL5000 series
On Thu, 20 Jan 2005, Jon LaBadie wrote: On Thu, Jan 20, 2005 at 10:53:26AM -0800, DK Smith wrote: It would be really cool if a single changer glue module was re-written in an OOP fashion, reusing the same logic code for every situation, but overriding the driver interface methods for specific driver interfaces. Ahh, you mean a changer-api to go along with the infamous DUMPER-API? Well... that _would_ actually be quite useful. But I'm not holding my breath waiting for it! :-) -Mitch
Re: Amanda and PostgreSQL
On Wed, 12 Jan 2005, [iso-8859-1] Germán C. Basisty wrote: 2. To use pg_dump to get a hot snapshot of the desired DB, put it on some directory and then backup that directory with amanda. We use pg_dump. We just run it from cron in advance of amanda's start time. -Mitch
Re: Amanda and PostgreSQL
On Thu, 13 Jan 2005, Jamie Wilkinson wrote: So I had a crazy idea that hasn't yet been implemented, a pg dumper similar in appearance to GNUTAR and DUMP on the client machine. Your DLE would look somethign like: host database/table pg where pg would be the dumptype telling amanda to use the pg dumper. The net effect is that you'd get a hot snapshot written straight to amanda, so you wouldn't have to worry about scheduling the dumps of each at the right time. If someone wants to beat me to the implementation though, please go right ahead :-) What you want is something called DUMPER-API. Unfortunately it doesn't actually exist. It's documented but still waiting for someone to code it. In the mean time, the standard workaround is to use the framework here: ftp://gandalf.cc.purdue.edu/pub/amanda/gtar-wrapper.* This is a wrapper script that will pretend to be gnutar to amanda, but run whatever you tell it to instead on the client. -Mitch
Re: Backing up directories with spaces
On Sat, 25 Dec 2004, Paul Bijnens wrote: I am trying to backup a directory with a space in its name, however, amanda doesn't seem to like it in the disklist. That's a known limitation of the current implementation. But there's an easy workaround: create a symlink on the client without spaces (if you can't rename the directory). That's only an easy workaround if you are an administrator of the client system. If you are not then you have to go to the include-style syntax. E.g.: host /VNTI /Library { nocomp-osx include ./Application?Support/VNTI?Database } The problem here is this is still buggy, as discussed on -hackers in April and July 2004. -Mitch
Re: amcheck not saying expecting tapeno. or a new tape
Would people please stop cross-posting between -users and -hackers. If your message is about using amanda, send it to -users. If it's about the source, send it to -hackers. Don't mean to single out Gavin here. This has happened several times recently, each then multiplied by all the followups. Thank you. -Mitch
Re: Is there a proper way of killing a dumper?
On Fri, 5 Nov 2004, Kevin Dalley wrote: I have one computer which is busy now, and is dumping *very* slowly. I want to kill the dumper for this computer. Killing all the dumpers is even OK. I could do a killall dumper, but it seems a bit crude. Is there a more polite way of killing a dumper, or just telling it to give up on a DLE? Don't kill the dumper, these get re-used from one client to the next. If you can get a shell on the slow client, go there and kill sendbackup or dump or tar. -Mitch
Re: auto-responder hell here
On Fri, 3 Sep 2004, Gene Heskett wrote: I consider such crap as exactly that, crap, totally useless crap at that, usually brought about by someones overblown ego trying to convince the rest of the world how important they are and without them the world will stop. Never ascribe to malice, that which can be explained by incompetence. -- Napoleon Bonaparte Or to put it differently, most of the morons who auto-reply to list mail these days really aren't doing it intentionally. The ones that annoy me enough I filter to /dev/null in procmail. Sometimes I send a nasty-gram instead. I rarely do both to the same sender. I guess I'm about to find out who today's brilliant 9 are... -Mitch
Re: sendbackup file renaming failure
On Thu, 26 Aug 2004, Paul Bijnens wrote: Mitch Collinsworth wrote: Today one of my dumps failed in an unusual manner. After the dump finished and the index was completed, I received the following error: error [renaming /var/adm/amanda/gnutar-lists/usda01afs:usda01_b_.*.backup_0.new to /var/adm/amanda/gnutar-lists/usda01afs:usda01_b_.*.backup_0: No such file or directory] I guess the '*' in the message above is a placeholder for the real name? Or is this the real filename? Even then, where is the '.backup' coming from? My gnutar-lists are named: hostname_disk_0 (with the last 0 indicating the level, and any '/' in the hostname or disk is replaced with a '_'). There is no place for a '.*.backup' in that scheme. The disklist entry is: usda01 afs:usda01/b/.*.backup nocomp-user So it follows the format you describe. The last 0 does indeed indicate a level 0 dump. Or does that has something to do with some afs modifications, as the name of the host suggests? The only modifications we made to amanda itself were 1) in selfcheck, to not try to test accessibility of a gnutar directory name beginning with afs, and 2) gnutar itself is replaced with a wrapper script that examines the DLE and chooses whether to run gnutar or an AFS command to produce the actual backup file. .*.backup is meaningful to the wrapper script. This has all been working fine for a couple of years now. There have been no recent changes. And despite this error on partition b, partitions a, c, and d ran just fine on this host. Somehow this was sufficient for the dump to fail. Therefore it was deleted from the holding disk rather than copied to tape. [Grr..!] The difference between success and failure of a backup is indeed a gray area, instead of a sharp line. This one is just on the boundary. The next dump would have trouble anyway because it wouldn't have a gnutar-list to base its incremental dumps on. That would result in a full dump, which is indeed a good fallback in that case. Certainly a better fallback than throwing up hands and putting nothing on tape! :-) In roughly 5 years of using amanda I have never seen this happen before. Me neither. This is 2.4.3b3. Try an upgrade to 2.4.4p3, at least on that client. Am planning to do this globally some time 'soon'. But it doesn't help with why this has been working fine for many months and suddenly messed up yesterday. There is one possibly-interesting thing about this one partition. It has been growing recently. Enough so that I have several times recently had to bump up dtimeout just for it. The problem there is that the program we substituted for doing indexing of AFS dumps seems particularly slow on large dumps containing many, many small files. I did have to do this just the day before this happened. Is it somehow possible that we ran into a corner case where it didn't timeout while the index was being created, but got so close that the timer ran out just after it finished and whoever checks the timer started some cleanup processing that zapped this file? And meanwhile whoever actually reports the timeout error did not do so, because the dump did in fact finish successfully? Sounds a bit crazy, but it's all I can think of so I may as well ask. :-) -Mitch
Re: Use of uninitialized value at /opt/amanda/sbin/amstatus line 868.
On Tue, 17 Aug 2004, Paul Bijnens wrote: Ranveer Attalia wrote: When running the amanda backup yesterday. It still doesnt appear to have finished. I checked the /tmp/ambackup_Daily.log file and its giving me the error: Use of uninitialized value at /opt/amanda/sbin/amstatus line 868 I tried very hard to reproduce the error, using the files you sent, but I'm unable to get any error like that. I don't see a response to this, but will add that I've been seeing something very similar. When I run amstatus I frequently (but not always!) see a few lines of this: Use of uninitialized value in numeric ne (!=) at /usr/local/amanda/sbin/amstatus line 653. I'm almost out of inspiration here. What version of perl are you using on the amanda server? /usr/bin/perl --version Here it's perl v5.6.1 and amanda 2.4.3b3. I'd been ignoring this, hoping it would go away when I upgrade to 2.4.4whatever, but since it's come up... -Mitch
Re: sendbackup file renaming failure
On Wed, 25 Aug 2004, Jim Summers wrote: On Wed, 2004-08-25 at 19:53, Mitch Collinsworth wrote: error [renaming /var/adm/amanda/gnutar-lists/usda01afs:usda01_b_.*.backup_0.new to /var/adm/amanda/gnutar-lists/usda01afs:usda01_b_.*.backup_0: No such file or directory] Just a guess, but possibly the fs involved has reached 100% capacity? Nope. -Mitch
Re: FreeBSD backups are extremely slow?
On Mon, 19 Jul 2004, Joe Rhett wrote: FYI, dumper appears to break the backup into 1gb chunks? I don't know if this is relevant or not. You mean on the holding disk? That's normal but you can configure the chunksize or turn it off if you wish. I'm seeing EXTREMELY slow backups of a moderately-sized file system on a FreeBSD host. I'm a little confused by this, as everything appears to be working .. it's just REALLY slow... The most common cause of this is a network duplex mismatch somewhere. Often between the host and the switch. Note that full---auto IS a mismatch. If one side is locked to full, the other side must be locked, too. -Mitch
Re: Rsync, Amanda's best friend
On Fri, 2 Jul 2004, Luc Lalonde wrote: In case of backup server failure, I 'rsync' all my configuration files in /etc with the Amanda index files onto another server. If my server crashes or becomes unavailable, I have instantaneaous access to my backup indices and configurations without mucking about and trying to find it on the backup tapes without an online Amanda setup. We keep ours in our OpenAFS filesystem. If our amanda server fails, we stand up another, fire up cfengine on it to install it as an amanda server, and off we go. -Mitch
Re: amanda and OpenAFS
On Fri, 25 Jun 2004, Frederic Medery wrote: I want to replace NFS by OpenAFS. I know that inside afs cell, root is not god anymore. So can I still use amanda with afs ? Hi Frederic, To save re-typing, here is a copy of my answer to someone who asked this a few weeks ago. -Mitch - From: Mitch Collinsworth [EMAIL PROTECTED] Date: Tue, 8 Jun 2004 16:38:38 -0400 (EDT) Subject: Re: amanda and OpenAFS? On Tue, 8 Jun 2004 [EMAIL PROTECTED] wrote: Is there anyone backing up OpenAFS using amanda? Is there anything special that needs to be done, or is it just like backing up any other file system? I'd like to use gnutar if possible. We are, and I've heard from a few other sites who also are. Don't have a good feel for how many yet. OpenAFS is not just like any other filesystem. In general, root is nobody special to OpenAFS. So your backup program needs to acquire appropriate privilege to read files. If you can do that (reauth is one way), you can just use gnutar. But then you'll back up the file contents without the ACLs and volume structure needed to put everything back the way it was before the Bad Thing that happened that caused your need to perform a restore. First stop: read docs/HOWTO-AFS This is just a pointer, but it tells you where to find some code we put together a few years ago to backup OpenAFS with amanda more or less right. I say or less because there are still some improvements I'd like to make sometime. One new item, for those who've heard all this before. Just today we created a new mailing list for amanda-afs users. Info can be found here: http://lists.ccmr.cornell.edu/mailman/listinfo/amanda-afs -Mitch
Re: amanda and OpenAFS?
On Tue, 8 Jun 2004 [EMAIL PROTECTED] wrote: Is there anyone backing up OpenAFS using amanda? Is there anything special that needs to be done, or is it just like backing up any other file system? I'd like to use gnutar if possible. We are, and I've heard from a few other sites who also are. Don't have a good feel for how many yet. OpenAFS is not just like any other filesystem. In general, root is nobody special to OpenAFS. So your backup program needs to acquire appropriate privilege to read files. If you can do that (reauth is one way), you can just use gnutar. But then you'll back up the file contents without the ACLs and volume structure needed to put everything back the way it was before the Bad Thing that happened that caused your need to perform a restore. First stop: read docs/HOWTO-AFS This is just a pointer, but it tells you where to find some code we put together a few years ago to backup OpenAFS with amanda more or less right. I say or less because there are still some improvements I'd like to make sometime. One new item, for those who've heard all this before. Just today we created a new mailing list for amanda-afs users. Info can be found here: http://lists.ccmr.cornell.edu/mailman/listinfo/amanda-afs -Mitch
Re: Managing out of the office twits
On Thu, 3 Jun 2004, Justin Gombos wrote: [anti_ooo.msg file] My scripts have detected that you posted an out of office reply to a public forum. Please control your auto-responder. If you are receiving this in error, I apologize; please disregard it. At the risk of prolonging this thread even further, it does seem incongruous for an auto-responder script aimed as mis-behaving auto-responders to request that if it itself mis-behaves, the recipient should please disregard. -Mitch
Re: Remote tape
On Tue, 27 Apr 2004, Paul Bijnens wrote: What you cannot do is have the amanda server drive a tapedrive on another host. Agreed, in the traditional sense. However iSCSI sounds like a means of accomplishing this. I say sounds because I haven't actually looked into it yet. -Mitch
Re: Amanda-induced sluggishness on Mac OS X?
On Sun, 15 Feb 2004, Kirk Strauser wrote: I use Amanda to backup an iMac running Mac OS X (10.2). The backup runs perfectly, but the machine is horribly unresponsive while tar is running, both during the estimate and dump stages. Renice'ing the tar processes doesn't seem to help, so I'm guessing that it might be an IDE DMA thing, although I don't know my way around Mac hardware well enough to know. Has anyone else seen this, and hopefully found a way to lessen the impact? Yes, we've seen it, too. Have not (yet) dug into it to try to figure out what's going on. The machine we've noticed it on is a Powerbook. -Mitch
Re: extremely varied tape write rates?
On Fri, 9 Jan 2004, Kurt Yoder wrote: In the report, see that some borneo tape writes were fast, but one is much slower. So why is that particluar one *always* slow? It should be just a mindless tape dump, right? Are you using a holding disk?
Re: amanda client on MacOS
On Tue, 6 Jan 2004, Paul Root wrote: it runs, the Mac times out waiting for a random local port. Any firewalls involved? -Mitch
Re: amanda client on MacOS
On Tue, 6 Jan 2004, Paul Root wrote: Nope. Same subnet Software firewall on client? Mitch Collinsworth wrote: On Tue, 6 Jan 2004, Paul Root wrote: it runs, the Mac times out waiting for a random local port. Any firewalls involved? -Mitch
Re: Estimates taking two hours - is this normal?
On Tue, 6 Jan 2004, Fran Fabrizio wrote: I have a system and I am attempting to backup a filesystem with approximately 30G of data. The estimates for a Level 0 take 3000 seconds. For a level 1, 7000 seconds. Is two hours just to get an estimate for an incremental dump? Is it typical to have to bump up the estimate timeouts to several hours? I just want to make sure that what I am seeing is normal. It seems like an awful lot of churning just to estimate a dump size. This happens to be my only Solaris system being backed up. On a linux system, I'm backing up an area approx. twice the size with no estimate timeouts. That may or may not have any relevance, but I thought it was worth mentioning. You might find estimates run faster with dump than tar. Which on a Sun is probably a safe thing to do, depending on what file system you use. -Mitch
RE: Dump Vs Tar tradeoffs (if any)
On Tue, 23 Dec 2003 [EMAIL PROTECTED] wrote: Also, I've found that 'tar' seems to take longer in the estimate phase than do 'dump' (and 'vxdump'), but for all I know, that could be due to local influences (e.g., disk traffic). Or has anyone else also found it to be the case that 'tar' estimates take longer than those from 'dump'? Tar has to stat every file to find out how much is going to be backed up. Dump just has to look at the filesystem as a whole. Far less work involved. -Mitch
Re: moved to new disk, now amanda wants to do level 0's on whole system
On Fri, 14 Nov 2003, Eric Siegerman wrote: But all of those -- tar, cpio, rsync -- are kludges. Is it just me, or do other people also find it ludicrous that 30+ years on, UNIX still doesn't have a proper copy command? Huh? You just showed there are enough flavors to suit just about any taste. The only obvious one I see missing is dd, which is great for copying a whole partition. -Mitch
Re: slow amanda performance on ONE system.
On Tue, 4 Nov 2003, Gene Heskett wrote: Good grief Toomas! Can you not institute a mail box size limit, about 10 megs maybe? Good grief indeed. This discussion has absolutely nothing to do with using amanda. Please take it offline. -Mitch P.S. To whomever posted the original 'one system too slow' query. Please double-check your duplex settings on the slow backing up machine. Mismatched duplex is one of the most common causes of this problem.
Re: Book: Automating UNIX and Linux Administration
The plug you sent to another list a few minutes ago ended with: This ends my one-time-only plug for my book. Thanks! Then you posted it again here, without the one-time-only caveat! So now I've received it twice, so far, and have to wonder how many more lists I will have to see it on before the end of the day! :-( -Mitch
Re: send req failed: Message too long
On Wed, 24 Sep 2003, Nicolas Ecarnot wrote: Nicolas Ecarnot a écrit : Hi, I recently had this kind of problem with my amanda server. It has to backup around 120 samba clients : GETTING ESTIMATES... planner: time 0.371: dgram_send_addr: sendto(192.168.10.66.10080) failed: Message too long send req failed: Message too long I'm sorry to insist but this problem is really blocking me and I don't really know what to do to solve it. I read in the faq that this may be due to an UDP packet size problem. The suggested workaround is to shorten the names used in the directories to save. Unfortunately, I'm backing up many samba clients, and their names can't be all changed, and I have no way to create symbolic links or that knid of things. In my disklist, I have 120 lines like this : backup.foo.com //winHost001/share nocomp-user-gnutar * I don't know HOW I could shorten this ? The other way to work around this problem is to employ multiple smbclient machines. So run some samba clients through backup.foo.com and some other samba clients through backup2.foo.com, etc. -Mitch
Re: About Linus's condemnation of Dump
On Sat, 6 Sep 2003, Scott Phelps wrote: Dump was a stupid program in the first place. Leave it behind. Linus has been bad-mouthing dump for as long as I can remember. Linus is obviously not a sysadmin or he would have more clue than this. Any comments, advice? (Does Amanda even work with xfsdump/xfsrestore? Works fine on IRIX. I haven't tried it on Linux, yet. Probably should. -Mitch
Re: Spam is getting old...
On Wed, 27 Aug 2003, Chris Barnes wrote: Changing an email address is a horrid way of dealing with spam. Either personally or for a list. *IF* it's a problem, running the email through a filter (my favorite is SpamAssassin) before it gets distributed is a MUCH better way to go. SpamAssassin has blocked all spam/virus mail arriving through this list for me since I started using it. (It hasn't done as well for non-list spam however.) So now I don't see the spam/virus, I just see the handful of 'you sent me a spam/virus' replies that follow each one and are sent erroneously to the list. If I were list manager I would either unzubscribe or block posting from sites generating these replies. -Mitch
Re: Spam is getting old...
On Wed, 27 Aug 2003, Kurt Yoder wrote: Mitch Collinsworth said: If I were list manager I would either unzubscribe or block posting from sites generating these replies. So who *is* the list manager who would be in charge of doing this anyway? I believe it's Todd Kover. Also note that I said If I were list manager. I'm not, Todd is. So it's his decision what to implement and what not. Not mine or anyone else's. -Mitch
Re: accessing data from tape on fresh amanda system (disaster recover)
On Mon, 11 Aug 2003, Jon LaBadie wrote: On Mon, Aug 11, 2003 at 11:15:08AM -0400, Mitch Collinsworth wrote: There's also amtoc. You can run it at the end of each amdump run. I like to then print its output to paper and save in a folder. When did that show up ??? It wasn't there the last time I looked :) Yah, I hate it when that happens, too. :-) From the comments in the script: # HISTORY # 1.0 19??-??-?? [EMAIL PROTECTED] # don't remember :-) # 2.0 1996-??-?? [EMAIL PROTECTED] # amanda 2.2.6 support # 3.0 1999-02-17 [EMAIL PROTECTED] # major rewrite, incompatible with release 2.0, amanda 2.4 support So it seems no one quite knows, but it wasn't recently. -Mitch
Re: accessing data from tape on fresh amanda system (disaster recover)
On Mon, 11 Aug 2003, Jon LaBadie wrote: On Mon, Aug 11, 2003 at 08:34:41AM -0600, Steven J. Backus wrote: Selon Jon LaBadie [EMAIL PROTECTED] writes; It helps greatly if there is a TOC of each tape available. Is there a slick way to create these rather than just print out the nightly email? I just set lbl-templ in my tapetype definition. In my case I use the 3hole.ps template provided with the distribution and it prints out on my default printer. There's also amtoc. You can run it at the end of each amdump run. I like to then print its output to paper and save in a folder.
Re: sdlt versus DLT library
On Tue, 15 Jul 2003, Kurt Yoder wrote: AFAIK it will hit end of tape and ask for another tape. However, there is no way to flush a single dump image to multiple tapes. So if you have a 20 GB dump image (even if it's been split up into pieces due to the chunksize parameter) and a 10 GB tape, there is no way to put the dump image on that tape. Does this strike anyone besides me as dumb? Yes I've known these details separately for years, but when they are put together in the same paragraph it jumps out and says Hey, the dump has already been split into managable-sized pieces. All we need is to modify taper to put them on multiple tapes and voila, we can span tapes! Of course then we'd need to do some fiddling with amrestore to be able to retrieve all the chunks during a restore... -Mitch
Re: OT vacation?
On Tue, 8 Jul 2003, Jeroen Heijungs wrote: Do I overlook something, or isn't there a way to temporarily disable delivery of mail during holidays? Most lists do have that possibility, but I cannot find it, it seems I have to unsubscribe and subscribe again. Whats the difference? -Mitch
Re: Help compiling on OS X/Darwin
One of my staff encountered the same problem with 2.4.4. He was able to successfully compile 2.4.3 and we've done some successful dump and restore tests with that. I was going to look into it further but haven't so far. -Mitch On Wed, 25 Jun 2003, Ken Simpson wrote: Hi, I am a newbie to UN*X, and have downloaded the 2.4.4 sources and read the notes on compiling on OS X and have searched the lists. However I can't get it to compile on OS X Jag. There are warnings in the configure and when I compile I get recursive errors in Amanda.h. Can anyone point me to a FAQ or a help page please? Or offer any other advice? Thanks -- Regards Ken Simpson .
Re: Using only one dumper
On Sun, 6 Apr 2003, Brandon D. Valentine wrote: You need to set maxdumps to something greater than 1 in your amanda.conf. It defaults to '1'. See amanda(8) for details. Urg. yes, maxdumps not inparallel. Somebody kick me for not checking my facts before posting. -Mitch
RE: Cygwin client - advantages?
On Sun, 23 Mar 2003, Bort, Paul wrote: 1. Client-side compression 2. Better chance of backing up ACLs (using special tar) 3. Works on machines that have all shares disabled (I have seen disabling shared recommended for some servers, especially SQL.) 4. More straightforward restore 5. Doesn't require SAMBA on AMANDA Server I haven't used it yet here, but I suspect another advantage not listed above is: 6. Can do dumps at higher levels than 0 and 1. Now my question: What special tar does one use to get the ACLs? -Mitch
Re: Pornography
Have we reached the point yet where the spam discussion has generated more list messages today than all the spam that's crossed it in the past year? -Mitch
Re: ACLs
On Mon, 10 Mar 2003, Adam Smith wrote: On FreeBSD 5.0 with UFS2 + ACLs, what is my best method for backing up my ACLs along with my files? I am only experimenting with Amanda at this point, but it seems to use the native tar utility, however tar does not support the backing up of ACLs. Can anyone show me what I need to do? I don't have a 5.0 system to look at, but looking briefly at the CVS tree, it appears their ACL system is designed to not require special consideration. There seems to be more explanation of the UFS1 implementation, which requires more manual setup, than the UFS2 implementation, which is natively available. Do your filesystems have a special directory at their root? Possibly .attribute ? (This is the name documented for UFS1.) If this is there then this is where your ACLs are stored and if you're backing up those dirs then you're getting the ACLs. -Mitch
Re: NAK: amandad busy
On Tue, 11 Feb 2003, Joshua Baker-LePain wrote: On Mon, 10 Feb 2003 at 5:06pm, justin m. clayton wrote I've been receiving this error on some, but not always all, of my hosts during amcheck. What could be causing this issue? Which logs are most likely to be housing the magic info I need to solve this? If there's already an amandad running (i.e. one from a previous run that never died), you'll get this error. Look in /tmp/amanda/amandad*debug. Also look at the output of 'ps' when you get the error. The other way this can happen is if you have multiple names in your disklist that refer to the same client machine. Amanda will consider them unique due to their names being different and try to contact each name separately, but the amanda client is not multi-threaded and will NAK the 2nd connection attempt if the 1st is still running. This should be a FAQ if it isn't already. -Mitch
RE: speed problem
Hi David, Sun hardware is really not my area of expertise, but I'm sure there are others on amanda-users who can answer this. If your Sun boxes will only do 10 Mbps/half-duplex then just make sure your switch ports are set to either auto/auto or 10/half and you'll be fine. -Mitch On Fri, 7 Feb 2003 [EMAIL PROTECTED] wrote: Mitch. Here is the version of my Sun OS: SunOS gitpocs02 5.7 Generic_106541-08 sun4u sparc SUNW,Ultra-5_10 The /dev/hme is FEPS Ethernet Driver v1.115 The host seems too old to support 100/full-duplex. Do you agree? Thanks! David -Original Message- From: Mitch Collinsworth [mailto:[EMAIL PROTECTED]] Sent: Thursday, February 06, 2003 7:18 PM To: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Subject: Re: speed problem On Thu, 6 Feb 2003 [EMAIL PROTECTED] wrote: My tape server is a Linux host (Redhat 7.3). My clients are either Linux hosts or Sun Solaris hosts. All client hosts are on the same domain. When I run 'amdump' to backup Linux hosts, the speed is pretty good. When I run 'amdump' to backup Sun Solaris hosts, the speed is extremely slow. It seems not a network or hardware issue. Seems? You don't sound particularly sure on this point... In my experience one thing that is a frequent cause of deadly slow network backups is a duplex mismatch between a backup client and the ethernet switch it connects to. I've seen systems that the user had been happily using without complaint for weeks or months before asking for backups. When their backup went painfully slow I'd check their network settings and invariably find a duplex mismatch. Another cause of slowness can be doing client compression, especially client best on slow hardware. How old/slow are your Solaris boxen? -Mitch
Re: speed problem
On Thu, 6 Feb 2003 [EMAIL PROTECTED] wrote: My tape server is a Linux host (Redhat 7.3). My clients are either Linux hosts or Sun Solaris hosts. All client hosts are on the same domain. When I run 'amdump' to backup Linux hosts, the speed is pretty good. When I run 'amdump' to backup Sun Solaris hosts, the speed is extremely slow. It seems not a network or hardware issue. Seems? You don't sound particularly sure on this point... In my experience one thing that is a frequent cause of deadly slow network backups is a duplex mismatch between a backup client and the ethernet switch it connects to. I've seen systems that the user had been happily using without complaint for weeks or months before asking for backups. When their backup went painfully slow I'd check their network settings and invariably find a duplex mismatch. Another cause of slowness can be doing client compression, especially client best on slow hardware. How old/slow are your Solaris boxen? -Mitch
Re: columm widths in daily mail report
On Thu, 30 Jan 2003, Joshua Baker-LePain wrote: On Thu, 30 Jan 2003 at 12:47pm, Don Carlton wrote Has anyone changed the mail report to produce html? Oh, the horror! ;) Agreed if sent by e-mail, but if it were instead dumped into a web-served directory... -Mitch
Re: TSM
On Wed, 15 Jan 2003, Aline wrote: Is TSM free? No. It's an IBM product... -Mitch
Re: advantages of amanda over ADSM or other backup utilities?
May I know some of the reasons why AMANDA edges over other traditional Unix backup utilities. stable does what it claims source available well supported good, unique scheduling module networked scales well from a single system to moderatly large installations Also that it uses native utilities for the backup images, so you can do a restore without having to reinstall Amanda first. This one is more important than it might first appear. In ADSM if you lose ADSM's database or it gets corrupted and you can't restore it, you can't restore anything else, even if the tapes your backups are on are perfectly fine. With amanda you can lose everything and still be able to restore from a tape with standard unix tools. Secondly, and I'm no ADSM/TSM expert, but the folks who run it here aren't yet offering OS X backups. I'm not sure if that's because the product doesn't support it yet or because of some other problem. Just this week someone reported here getting amanda to do OS X backups ok. But it should be noted that there are advantages of ADSM/TSM as well. One is that the client can request a backup asyncronously. Handy for e.g. the laptop user who is only connected sporadically. People have hacked up various workarounds for this for amanda, but it doesn't do it out of the box, yet. Another is that ADSM/TSM recognizes a mobile machine whose IP is changing. If you backup a laptop at work, then take it home or wherever and request a fresh backup, it does the right thing. Amanda would require a bit of work to do this. Another is their incrementals forever. I've argued this one both ways, but it can be handy in some situations. ADSM/TSM takes one full dump when the client is first backed up and then does only incrementals from there on. They can do this because they backup files rather than filesystems, and keep track of everything in the previously-mentioned database. The big advantage to this is when you want to put a machine behind a slow, overcommitted, or expensive network link. You could for example do a full dump on-site first, then move the machine to the remote location with the slow link and it will only need to do incremenals from there on. -Mitch
Re: amandad busy
On Mon, 30 Dec 2002, Adnan Olia wrote: I was just wondering why would amanda send a message amandad busy. Do I need to check before running amdump that would solve my problem?? Typically this means that a previous run has not yet finished. Maybe something got hung? Does ps show any amanda processes running? Another possibility is that you listed multiple filesystems from the same client with different hostnames in disklist. This would cause amanda server to view them as different clients and attempt a separate connection to each. The second connection will find amandad already busy from the first connection. -Mitch
Re: Problem with initial install of amanda on SCO openserver
On Mon, 30 Dec 2002, Josh More wrote: FAILURE AND STRANGE DUMP SUMMARY: backupmast /home RESULTS MISSING Firstly, before running amdump did you run amcheck? If yes, what do you see in /tmp/amanda? -Mitch
Re: Still get No index records...
On Mon, 30 Dec 2002, John Oliver wrote: 501 No index records for host: backup.indyme.local. Invalid? Trying backup.indyme.local ... 501 No index records for host: backup.indyme.local. Invalid? Trying backup ... 501 No index records for host: backup. Invalid? What did you use for the hostname for this client in disklist? -Mitch
Re: COMPAQ TSL-9000 DAT autoloader device
On Sat, 23 Nov 2002, Jon LaBadie wrote: Often the same device is sold under many brand name plates. I don't think Sun manufactures any tape drives and Sears doesn't make any dishwashers. They just relabel/repackage others products. And often a library/changer sold by one company uses a standard tape drive from another company. All tapetype cares about is the drive, not the changer. Yes. But often that standard tape drive has custom firmware installed for use in the library/changer. Which makes it non-standard in terms of ordering and pricing. -Mitch
Re: use --build, --host, --target
On Tue, 24 Sep 2002, John Dalbec wrote: Hadad wrote: [root@localhost amanda-2.4.3b4]# ./configure --disable-libtool --without-client --with-user=amanda and --with-group=amanda Take out the word and here and you should be OK. That is, ./configure --disable-libtool --without-client --with-user=amanda \ --with-group=amanda I think he also doesn't want --without-client. Didn't he say hes going to back up just this one machine? It will need the client pieces installed in order for that to work. -Mitch
Re: backing up commercial apps
On Sat, 21 Sep 2002, Neil wrote: What is amanda's approach to backup commercial software like Microsoft Exchange 5.5 mail server? It's because in Exchange, they have this public Jon's reply is a good description of amanda itself. Haven't checked into it yet but I've heard there's another project at sourceforge that's working on a native windows client for amanda. It may well be better than doing the samba thing. Will there be any future path that Amanda is looking into? Or is Amanda is just really looking into perfecting back up of unix filesystem? I also heard that backing up registry of the Windows Oses is a bit crappy at the moment. Well as Jon pointed out amanda isn't perfecting any filesystem's backup. That's the filesystem maintainer's job. Amanda just wants to make it easier for you to run backups so you can spend more time doing something else. Those of us wanting to backup fs's and db's that amanda doesn't already do are building bridges between amanda's architecture and the backup tools used by the things we want to backup. Registry backup is a challenge for most windows backup schemes. There is a tool somewhere (sorry, don't recall the name) that will snapshot the registry into a quiescent file. If you want to get a good backup of the registry, find this tool and schedule it to run before your backup starts. Then backup the registry dump file. This is pretty much the same scheme used to backup most any database regardless of OS. -Mitch
Re: Deja vu all over again
Here's the relevent section of headers: Received: from pwd.reeusda.gov ([192.73.224.125]) by surly.omniscient.com (8.11.6/8.11.6) with ESMTP id g8IImln797017 for [EMAIL PROTECTED]; Wed, 18 Sep 2002 14:48:47 -0400 (EDT) Received: from reeusda.gov - 192.73.224.2 by pwd.reeusda.gov with Microsoft SMTPSVC(5.5.1774.114.11); Wed, 18 Sep 2002 14:46:22 -0400 Received: (qmail 17216 invoked by uid 0); 18 Sep 2002 18:32:14 - Received: from guiness.omniscient.com (64.134.101.78) by maat.reeusda.gov with SMTP; 11 Sep 2002 18:30:49 - -Mitch On Wed, 18 Sep 2002, Frank Smith wrote: What's up with last week's postings getting re-injected into the list? Frank -- Frank Smith[EMAIL PROTECTED] Systems Administrator Voice: 512-374-4673 Hoover's Online Fax: 512-374-4501
Re: Configuration help?
On Fri, 16 Aug 2002, Joshua Baker-LePain wrote: Tread lightly -- thar be dragons here. :) They both have their pros and cons, and it can be a matter of deep seated religous belief for people. FS specific dump programs can back up things that tar doesn't know about (e.g. ACLs), and can sometimes be faster. But they're limited to partitions only, and require that the recovery machine have them installed. Tar can do subdirectories and doesn't care about OS/FS (and thus you can recover on just about any machine). It's all a matter of choice. Basically I agree with Joshua, with one addition. For Linux systems stick with tar. dump on Linux is, well, not know for being robust. -Mitch
Re: Holding Disk Question
On Thu, 15 Aug 2002, Kevin Passey wrote: What is the best size for a holding disk - is it a how long is a piece string or are there some optimal values. It mostly a tuning thing that's up to you to decide what's best for you. That said, I prefer my setups to have as a minimum, enough to hold any single night's backup in its entirety. Then if something bad happens with the tape drive/media, the backups can run to completion spooled to the holding disk(s) and in the morning I can solve the tape problem and flush the holding disk(s) to tape then. Also, having enough to hold all of a single run allows you to take the greatest advantage of dumping parallelism, thus getting your client backups finished as quickly as possible. Further improvements on this theme are, 1) enough holding disk to hold an entire weekend's backups, and 2) enough holding disk to hold an entire insert length of your longest anticipated family vacation here. My motto: Disk is cheap, don't skimp on holding disk. -Mitch
Re: Holding Disk Question
On Thu, 15 Aug 2002, John Koenig wrote: Yup... after I deployed a 181 GB drive for our holding disk, I saw somewhat different behavior in amstatus... though I would have expected the total time of the backups to drop (compared to a relatively small 25 GB holding space I used before) they did not... I conclude this was because the entire run is tape i/o bound... Holding disk usage on the runs with this disk were about 90% capacity... so the clients were not taking so long to complete their data transfers... This would seem to indicate I need to double this amount of space to hold 2 days of backups, currently... I'd like to have a week's worth, actually in case the tape goes South and replacement is not easy and quick. So does the chunksize parameter affect performance in any way? No. Taping of a dump spooled to holding disk does not begin until all chunks are received. The best way to figure out where your bottleneck and what configuration changes will give you the greatest return is is to run amplot. It will give you a graphical representation of what's happening when. Very revealing. -Mitch
Re: Holding Disk Question
On Thu, 15 Aug 2002, Gene Heskett wrote: On Thursday 15 August 2002 14:49, John Koenig wrote: So does the chunksize parameter affect performance in any way? Only in that it prevents troubles with a filesystem that can't handle large files. There was, at the time amanda was first deployed, a 2gig limit to the filesize in many of its various platforms filesystems. Many of those limits are now historical, but you'd hate to find it out by doing a recovery and having it blow up because a tape of 20gigs was all one big file. A 20 GB dump _is_ all one big file. Chunking is only implemented on the holding disk during dumping. The chunks are merged back into a single file when spooled from holding disk to tape. When you do a recovery what comes back is a single large file. If you want to recover the whole dump from tape to your holding disk before pulling out the files that need to be restored, then you'll have to split it up by hand in the process. -Mitch
Re: No end to problems
On Sat, 3 Aug 2002, Paul G. Allen wrote: [root@ista root]# amtape csd show amtape: scanning all 12 slots in tape-changer rack: slot 2: reading label: Input/output error slot 3: reading label: Input/output error slot 4: reading label: Input/output error slot 5: reading label: Input/output error slot 6: reading label: Input/output error slot 7: reading label: Input/output error slot 8: reading label: Input/output error slot 9: reading label: Input/output error slot 10: reading label: Input/output error slot 11: reading label: Input/output error slot 0: reading label: Input/output error One common problem is that some changers will return from a tape load request before the tape is fully loaded and the tape drive is ready to go to work. In these cases it becomes necessary to add some delay time between the tape load command and subsequent tape drive use commands, or else do a sleep/retry/timeout sequence. Here's what worked for me a few years back with a DLT4000 changer: # see if tape drive is ready. if not, sleep a while and check again $remaining = $MAXDELAY; while (!open(TAPE,$TAPE) $remaining 0) { $remaining -= sleep 5; } if ($remaining = 0) { print $0 drive $TAPE timed out, not responding\n; exit 2; } I did this just after loading a tape, so any following commands would not have to worry about it. For that changer it took I think close to a minute for the drive to become ready. More recently, with an AIT2 changer I found the open() call here didn't work, so replaced it with a call to mt status and watched the output for General status bits on. Whatever works. :-) -Mitch
Re: amanda report
Hmm... Looks like you didn't have a (usable) tape in the drive. Were the level 0 dumps of these new disks bigger than would fit on your holding disk(s)? -Mitch On Wed, 31 Jul 2002, Michael P. Blinn wrote: A few questions about my report. I have a cron set to force level 0 dumps a minute or two before the actual backup goes into place so that I can get a full level 0 on each tape. I understand can't switch to incremental dump because it was forced to be dump 0, right? But why would the entire backup fail for this one machine? It is on and available! Thanks in advance, -Michael Blinn FAILURE AND STRANGE DUMP SUMMARY: mail /configs lev 0 FAILED [can't switch to incremental dump] mail /var/lib/mysql lev 0 FAILED [can't switch to incremental dump] mail /usr/local/apache lev 0 FAILED [can't switch to incremental dump] STATISTICS: Total Full Daily Estimate Time (hrs:min)0:00 Run Time (hrs:min) 0:12 Dump Time (hrs:min)0:12 0:00 0:12 Output Size (meg) 158.60.0 158.6 Original Size (meg) 607.40.0 607.4 Avg Compressed Size (%)26.1--26.1 (level:#disks ...) Filesystems Dumped2 0 2 (1:2) Avg Dump Rate (k/s) 221.6-- 221.6 Tape Time (hrs:min)0:00 0:00 0:00 Tape Size (meg) 0.00.00.0 Tape Used (%) 0.00.00.0 Filesystems Taped 0 0 0 Avg Tp Write Rate (k/s) -- -- -- NOTES: planner: Adding new disk mail:/usr/local/apache. planner: Adding new disk mail:/var/lib/mysql. planner: Adding new disk mail:/configs. planner: Last full dump of localhost://ntserver/ppidocs on tape daily5 overwritten in 2 runs. DUMP SUMMARY: DUMPER STATSTAPER STATS HOSTNAME DISKL ORIG-KB OUT-KB COMP% MMM:SS KB/s MMM:SS KB/s -- - localhost-sinesswork 1 19320 3968 20.5 0:54 73.7 N/A N/A localhost-er/ppidocs 1 602623 158400 26.3 11:19 233.3 N/A N/A mail /configs0 FAILED --- mail -cal/apache 0 FAILED --- mail -/lib/mysql 0 FAILED --- (brought to you by Amanda version 2.4.2p2)
Re: Survey of Changer Users: After amdump returns, where is tape?
On Sun, 21 Jul 2002, Gene Heskett wrote: On Sunday 21 July 2002 02:44, Mitch Collinsworth wrote: You asked what amanda does with changers. I believe when amdump completes the tape is rewound but not ejected. Its not even rewound here Mitch. Since amanda uses the non-rewinding device, the only time a rewind command is issued is when it wants to verify its got the right tape to write to. amcheck does this, and amdump does at startup, but not again. Sorry. I should be more careful about what I believe at 2:30 AM. I've used a number of different tape types and not seen any problem that could be traced to leaving a tape in a drive. But Exabyte is a drive I've had very little experience with. I can verify that too, I've had zilch trouble from leaving the tape in the drive, using a Seagate 4586np changer with a 4 tape magazine. FWIW, once one of these changers is loaded, I found that an 'mt -f /dev/nst0 rewoffl' will in fact cause it to load the next tape after that one has been rejected. Interesting. I've seen different behavior from different changers, but I haven't seen this. Is this in gravity mode? -Mitch
Re: Hardware to backup upto 60-80 GB
On Fri, 19 Jul 2002, Marcus Schopen wrote: anyone knows a good Streamer to backup 60-80 GB with Amanda? AIT-2, AIT-3, DLT7000, DLT8000, LTO -Mitch
Re: Times from amreport
On Fri, 19 Jul 2002, Ulrik Sandberg wrote: Estimate Time (hrs:min)1:25 Run Time (hrs:min) 4:02 Dump Time (hrs:min)3:07 2:38 0:30 Tape Time (hrs:min)0:43 0:42 0:02 Does this mean: a) Total run time (estimate, dump and tape): 4:02 b) Total run time (estimate, dump and tape): 5:27 (4:02 + 1:25) Total run time is 4:02, as stated. The numbers don't add up because there is parallelism to be had from dumping simultaneous dumps to holding disk, plus there can be another dump simultaneously being spooled to from holding disk to tape. -Mitch
Re: Windows-XP
On Thu, 18 Jul 2002, Gene Heskett wrote: On Thursday 18 July 2002 12:33, Marc Mengel wrote: Amanda-hackers -- shouldn't we have a link to their project on the main Amanda page, next to the SAMBA links? Marc Much as we'ed like to be a bit political here, this being primarily a *x type list, I have to agree with Marc. Only by inviting the windows users in can we make it 100% compatible on a faster timetable, and widen the userbase of amanda. I strongly agree. Good tape drives and tapes are expensive. There is a strong incentive to find one backup system that will handle all your backup needs. The samba stuff sorta-kinda works, but having a good client that runs directly on M$ os's will enable a lot more folks to choose amanda. -Mitch
Re: amanda in an Andrew environment?
We've been working on amanda backups of AFS here, but not using AFS's backup system. We found that too painful. We ended up building our own dump program that wraps around vos dump. It's often said that DUMPER-API is coming, but I've seen nothing to suggest there's been any recent progress. Some preliminary work was done on it a while back but I think it stopped. I know very little about Coda but I don't see a need for an AFS backup that produces BSD dump-format output. We also wanted to be able to restore AFS dumps into non-AFS filesystems, but we were able to build a restore program that does this. It's just a matter of being able to decode the vos dump format. -Mitch On 11 Jul 2002, Greg Troxel wrote: Search the archives and google. Previous posters have mentioned modified amanda versions that incorporate the afs dump tool. I am only familiar with Coda (www.coda.cs.cmu.edu), which has its own backup program. Basically the idea is to use amanda to transport and store the bytestream from the native 'dump' version for the non-ufs filesystem. For various reasons, I use gnu tar to back up Coda. This loses the atomic snapshot that the 'real' backup does, and doesn't back up coda metadata (acls), but it works well enough. With the coming 'DUMPER' API, it should be possible to hook in native coda and afs clone/dump programs. I believe the coda folks have a modified amanda that does Coda. rant I really wish that the coda and afs backup programs would produce a BSD dump-format stream, with the files as files, and metadata as files with some funny names (e.g. lots of __, or ^A^F^S), so that one could restore them either on the coda/afs filesystems, or read them anywhere else. But this is really orthogonal to amanda running sendbackup-coda.c. /rant Greg Troxel [EMAIL PROTECTED]
Re: New Amanda release (off-topic)
On Mon, 8 Jul 2002, Jonathan R. Johnson wrote: * * * * New Release of Amanda * * * * * Amanda Ruth Johnson * * Born June 24, 2002 24 June, 2002 * 8 lbs, 2 oz(translated out 3.7 kg * 20 in of American) - 50.8 cm Congratulations! And born on my birthday, too! :-) -Mitch
Re: syncronize backups
On Wed, 3 Jul 2002, Scaglione Ermanno wrote: This is a CISCO document explaining NAT: http://www.cisco.com/warp/public/cc/pd/iosw/ioft/ionetn/prodlit/1195_pp.htm It states that Any TCP/UDP traffic that does not carry source and/or destination IP addresses in the application data stream is supported and without the sendsize problem also amanda UDP traffic is supported. When port translation is configured, there is finer control over translation entry timeouts, because each entry contains more context about the traffic using it. Non-DNS UDP translations time out after 5 minutes; DNS times out in 1 minute. TCP translations time out after 24 hours, unless a RST or FIN is seen on the stream, in which case it times out in 1 minute. The problem exists certainly also with linux firewall using iptables becouse it uses even smaller timeouts. Worse yet, iptables in Red Hat 7.x doesn't allow control of the timeout value. A huge step backwards from ipchains in Red Hat 6.x, which did allow tuning this parameter. Amanda is not the only thing that breaks without this feature. -Mitch