Re: Laptop Backup Strategy
I'm beginning to investigate using external USB 2.0 / Firewire drives to do backups of some systems. The idea is that we could keep 2-3 spare PCs around, then if your computer is toast, we just ship it out for repairs, plug your external drive into one of the spare PCs, and rebuild the system from the external drive. --jonathan
Re: Looking for tape drive suggestions
I'm going to work for the Protein Data Bank, and we're seriously talking about using 1 TB flash drives for backups in the not-too-distant future. It may take a few years to get down to a reasonable price point, however. --jonathan
Bacula vs. amanda?
I sent this From the wrong address initially, apologies if you actually get it twice :( My experience is with amanda, but I will be taking over a Bacula installation in about a month when I change jobs, see http://www.bacula.org/ Has anybody used Bacula and have any comments on how it compares to amanda? I'll also have a bit more budget to play with in my new role, so might consider just changing to a commercial backup system, but I'll also be handling much fewer machines, which will greatly simplify the ongoing maintenance of amanda anyway. Thanks, --jonathan
Re: Restoring an ArcServe tape on Linux without ArcServe
Martin Hepworth wrote: dd the entire tape to disk, so you've got the binary to play with^W^W inspect. Exactly what I was thinking. Next, I'd probably use split to break it down into manageable chunks and use strings and od commands such as od -c and hex editors to try to make some sense of the junk. With luck, you might find some pattern, like the beginning/end of individual backups or files that you could use csplit to break the backup into more logical chunks. With even more luck, you might figure out what compression algorithm was used, for example gzip compressed data almost always begins with the same character string, but I forget what it is, then you might be able to decompress the chunks or something if it isn't totally proprietary. --jonathan
Re: Restoring an ArcServe tape on Linux without ArcServe
Jon LaBadie wrote: On Wed, Sep 08, 2004 at 11:57:37AM -0400, Jonathan Dill wrote: ... . With even more luck, you might figure out what compression algorithm was used, for example gzip compressed data almost always begins with the same character string, but I forget what it is, See /etc/magic for the list of magic numbers the file(1) uses. On Linux, that's often burried in /usr/share somewhere these days, for example I just checked on Mandrake 10 it's /usr/share/misc/file/magic, rpm -q -l file will help on RPM-based systems, or locate (from slocate if installed) might help find it.
Re: DVD Arrays (was Re: Is anyone using a dvd drive yet?)
Good points. For small backups, I think that removable firewire / USB 2.0 drives would be a more economical and convenient option, using FILE-DRIVER to write vtapes onto the removable drives. I still think DVDs are more suited to archival backups of a few GB where you plan to store and delete the data. Lately, I keep getting the feeling like people want a $50k backup solution but only want to pay $3k to get it and then complain when it doesn't live up to their expectations. Frank Smith wrote: At one time I thought DVDs would be a good backup media, but I have since changed my mind. The main reason is capacity (or lack of it). Unless you are backing up small amounts of data, it would take stacks of DVDs to perform your backups. Perhaps now that dual-layer burners and the new Blu-ray format are emerging the pain threshold is rising, but currently I cant see using 25 DVDs instead of a single 100 GB tape (and those aren't even the largest tapes commonly available). If your needs can be handled by a few DVDs, the easiest way to implement it might be to use Amanda's file driver and then burn that to DVD. Frank
restore from multiple holding disk files
Hi folks, I'm trying to help someone do a restore from a dump that is split into multiple chunks in holding disk files. In this case, flushing to tape first is not an option. I thought amrestore could do it, but then I read the manpage and didn't see a way to do it. The only way that I could think of doing it is to use dd to strip the amanda header off the chunks and then concatenate them together into one, huge file. The filesystem on the holding disk is XFS so the filesize should not be a problem. Is there an easier way to do this? The files are ufsdump being restored to a Solaris machine, but the holding disk files are on a Linux box which does not support ufsrestore. Thanks, --jonathan
Re: restore from multiple holding disk files
Paul Bijnens wrote: If split in chunks, just feed the first part; the rest is done automatically (the name of the next part is in the header of each holding chunk). Hmm. That's what I thought, finding subsequent chunks might not be working correctly then for whatever reason, but I'll have to search for the e-mail from the person who was having the problem. --jonathan
Re: backing up amanda server
On Tue, 2004-05-18 at 10:14, Joe Konecny wrote: Am I correct in thinking that if I am backing up only the server running amanda then it makes no sense to use the holding disk? Not exactly. If you are backing up files from the holding disk, for example in other directories, then it probably makes no sense to use the holding disk. For filesystems other than the holding disk, you may still get better tape performance by using the holding disk because dumps may stream to tape better than dumping directly to tape. Offhand, I'd say use the holding disk. If you really care, then try it without the holding disk and compare the statistics with and without using the holding disk. In a nutshell, the 2 questions are: 1) How fast does your tapedrive gobble up data? 2) How fast can the backup process feed data to the buffer(s)? The goal is to keep the tapedrive streaming as much as possible, which means keeping a full buffer, which means your backup process better generate data faster than the tapedrive can absorb it. Otherwise, every time the buffer goes empty, the tapedrive has to stop and reposition, which takes time and may eat up space on the tape, making for slower backups that use up more tape to store the same amount of data. If the amanda server is the only client, I'd probably turn on hardware compression on the tapedrive, turn off software compression, and maybe dump directly to tape. If there are other clients involved, and you want the bandwidth savings of software compression client-side (which I do) then I'd stick to software compression for everything and use the holding disk on the amanda server, since the software compression would slow down the backup process and increase the chance of emptying the buffer. Why then use software compression for everything? I'm not sure if it's a good idea, or necessarily possible, to mix hardware compressed and uncompressed data on the same tape, or to get amanda to somehow switch the tapedrive between those modes. I am sure that applying hardware compression to data that is already software compressed will use at least as much, and probably more (and possibly a lot more) tape. -- Jonathan Dill [EMAIL PROTECTED]
Re: backing up amanda server
On Tue, 2004-05-18 at 10:08, Paul Bijnens wrote: And there are the cases when there are problems with the tapedrive (wrong tape, holiday period, operator couldn't make it in the office because he's sick etc.) Having (at least incremental) backups on holdingdisk is a nice fallback. Oh yeah that reminds me...If your average amanda runs are much smaller than the size of a tape, then you might consider a strategy of usually backing up to holding disk, and periodically running amflush to try to get the tapes as full as possible. I have one amanda setup that I run that way. -- Jonathan Dill [EMAIL PROTECTED]
Re: Max File Size
The limitation could be that the login shell is not compiled with LARGEFILE enabled--that was a bug that plagued tcsh in various distros for awhile, though I haven't seen that bug lately. In the past, I ended up getting the Mandrake SRPM for tcsh and porting it into Red Hat to fix the problem, or just using bash. I think there's a way to use strace or strings to see if LARGEFILE was enabled when the shell was compiled, but it's been so long I don't remember--maybe someone else can chime in with the answer, or you could dredge it up from the list archives. On Fri, 2004-05-14 at 17:53, Josh Welch wrote: I am thinking that things are dying as a result of this friggin huge file, I am able to restore just fine from a smaller backup on a different machine. I -- Jonathan Dill [EMAIL PROTECTED]
Re: location of amandahosts
On Mon, 2004-05-17 at 16:01, Eric Siegerman wrote: On Mon, May 17, 2004 at 03:40:16PM -0400, Joe Konecny wrote: First install of amanda... Freebsd 5.2.1, Amanda 2.4.4p2. I used bin and operator when compiling. I much prefer to create a new userid just for Amanda. If it runs as bin, then it can write to a large part of the system (no For compatibility with pre-compiled RPMs for Linux, I like to use amanda UID 33 and disk GID 6 that just keeps everything consistent across all of the platforms that I back up. GID 6 has read but not write access to the disk devices for dump style backups (xfsdump, ufsdump, dump etc.) and no users are members of that group. -- Jonathan Dill [EMAIL PROTECTED]
Re: your mail
Gavin Henry wrote: Sorry, forgot to put in a subject. Yes, Fedora. Is there anything else like it for Linux? I have periodically searched for ufsrestore-compatible software for Linux, but have been unable to find any yet. In my case, I wanted to be able to index dumps from a Solaris amanda client on a Linux amanda server. A few ideas, which might be useful for the future at least: 1) switch to GNUTAR (I was using GNUTAR but had ended up switching to ufsdump due to some problems); 2) there used to be a binary compatibility layer that let you execute Solaris executables on Linux, but maybe that has disappeared, or at least would probably require compiling a custom kernel; 3) look for source code for ufsrestore, but I don't think it's Open Source; 4) Can Solaris run under VMware on x86 architecture? 5) look for Open Source libraries that can parse ufsdump format, or ufs filesystem, and write your own code. If #2 still exists (though I don't think so) there might be a Linux distro out there tailored to exploit that capability. --jonathan
Re: FW: unsubscribe please
As a stopgap, perhaps there is a way to turn off receipt of mail? On LISTSERV it was something like set nomail Technically, you would still be subscribed, but you would not receive any messages. On Tue, 2004-05-11 at 08:57, Bryan K. Walton wrote: I'm another person who was magically re-subscribed yesterday. Is there someone on this list who and explain what has happened. Also, I'm not able to unsubscribe either. Getting an error message back from Majordomo. -- Jonathan Dill [EMAIL PROTECTED]
Re: unsubscribe
OK just for fun, I gave it a shot, and it looks to me like someone used RCS (ci, co, rcsdiff etc.) to manage something within the Majordomo tree and Majordomo is unhappy about having an RCS directory within the tree because it is not a valid list. That's what I suspect. You're probably going to have to: 1) store the *,v files in another directory; 2) remove the RCS directory; 3) or, figure out how to make Majordomo ignore the RCS directory. Why oh why do people still insist on using antiquated Majordomo when there are a dozen or so better MLMs out there these days? Try Mailman for instance, available as an RPM on multiple Linux distros. -- unsubscribe * [EMAIL PROTECTED] unsubscribe: unknown list 'RCS'. Your request to [EMAIL PROTECTED]: -- Jonathan Dill [EMAIL PROTECTED]
Re: New to Amanda- discouraged by some absurd limitations..
Which reminds me...If cost is a factor, now that FILE-DRIVER is an option, RAID or removable hard drives may give you a better $/GB ratio than tapes, and much more capacity than CD-R. I think this is a very good option for a single computer or small network like Justin described in his original e-mail. If you use removable drives or a RAID-1, you might not need anything else, though it would be a good idea to still dump more important files to tape or writable DVD media occasionally. I haven't investigated it, but I have heard of hot-swap external SATA and firewire options which would be very good indeed. 250 GB removable drives could be a great option if you are backing up large partitions, say up to 500 GB uncompressed, so that you could get around dumps not fitting on a single tape without having to use RAIT with multiple tape drives, or very expensive tape drives and media, or split up dumps with (IMHO inefficient and CPU/IO intensive) GNUTAR. In my case, I am using a 1 TB Snap Server 4500 in a RAID-5 configuration and flushing mostly just the full dumps to 200/100 GB LTO (Ultrium-1). Since RAID-5 has less redundancy than RAID-1, I am more concerned about having at least some dumps on tape since a 2-disk failure would mean that all of the data on the RAID-5 would be gone. On Sun, 2004-05-02 at 20:53, [EMAIL PROTECTED] wrote: In my Amanda experience, I was lucky enough to have a large holding disk area and a tape drive which failed spectacularly before even one backup was flushed. It gave me the opportunity to see how Amanda works. The most wonderful aspect was how happy she was to restore from the holding disk. --jonathan
Re: New to Amanda- discouraged by some absurd limitations..
Jon LaBadie wrote: One of the concerns I have about disk-only based backup schemes is the total loss of data. If you encounter a 2-disk failure you lose not only your most recent, but all your backups. If a tape drive fails the data can be read on another drive. If a single tape goes bad, that is the only set of backups lost. If you're using multiple removable hard drives as tapes, I think that would mitigate the risk somewhat vs. disks that are online all the time, although spin-up seems to be THE most crucial moment in the life of any disk drive. I think disks still aren't as reliable as tapes in terms of failure rates, but would expect a removable hard drive solution to fall somewhere between full-time disk drives and tapes. --jonathan
Re: New to Amanda- discouraged by some absurd limitations..
Use FILE-DRIVER and wait until you have enough files to fill up the CD-R. Or use CD-RW as your tapes and keep several that you can rotate and re-use. Oh yes, we have designed amanda specifically to satisfy your personal whims, pretty please don't reject it, it will so much hurt my feelings. IMHO this was rather rude way to bring up the issue, Why not just look for some other software if you didn't like it? No one held a gun to your head and made you use amanda. freshmeat.net lists several other packages specifically geared to making backups to CD-R as you describe. amanda is geared to backing up large networks, like Veritas or Legato without the very expensive licenses. You can get it to work for a small, single computer, but it was not designed to do that, hence it may well not be the best tool for that job, nor does it promise to be. Justin Gombos wrote: I was looking forward to using Amanda to backup 4-5 machines on my LAN and one over the Internet, but something seems incredibly stupid about the way Amanda forces the user to operate. Please tell me I'm wrong; maybe I'm misunderstanding the documentation. If I want to perform daily backups to CDRs, and I expect to have around 10 megs of data change per day, do I really have to waste an entire CDR every day? --jonathan
Re: New to Amanda- discouraged by some absurd limitations..
Justin Gombos wrote: First of all, I did not know whether this was an issue, that's why I posted here. It was certainly proper to raise an Amanda issue/question to the amanda-users mailing list. Asking, Can amanda do X? is one thing, but to complain of an absurd limitation is, frankly, insulting, unless that was a bad translation of something from another language. OTOH-- if you want to talk about etiquette, you should try not to ignore the mail-followup-to header on mailing list posts. I don't see how the use of insulting language is equal to whether my e-mail client correctly handles mail-followup-to headers. If you know of a multi-client GNU backup tool that both works over the network and is also uses the target media intelligently, please advise. Otherwise, someone might as well be holding a gun to my head forcing me to use Amanda, because it seems to be the closest tool for meeting this requirement. The other tools I've studied lack the automated across network ability. As others have pointed out, 700 MB CD-R is probably not going to be adequate for backing up multiple clients across the network anyway. By GNU do you mean covered under the terms of the Gnu Public License? Are you opposed to other free, or free to use but restricted licensing? Is the point that you want something free or very cheap, hence would not consider something like Arkeia? --jonathan
Re: star and amanda
Amanda doesn't use tar to write to the tape directly, so the ability of tar or star to span tapes or not isn't the issue. Amanda makes a dump image, which has a header followed by what ever backup format was used on the client OS. Then the image is then written out to tape, even if you are backing up direct to tape through amanda. Besides, I swear that I can remember doing tar to multiple floppies with regular old GNU tar by specifying a volume size and multiple flag or something like that. The issue with amanda is whether or not you can span the dump images across multiple tapes. I thought there had been some progress in that area, or at least some possible hacks using FILE-DRIVER or something, though it still might not be completely straightforward. In freshmeat.net the other day I think I noticed there were some other GUI front-ends to amanda out there, but I have not checked them out. Also, Red Hat is including, and supposedly supporting, amanda on Red Hat Enterprise, and may have made some advances in the configuration area. I think a lot of GUI for backups tend to be a bit clunky and mask a little too much of what they are actually doing in the background, but then I am still a bit of a console cowboy. Windows in general can drive me crazy that way sometimes, when something is broken and there is no easy, free way to get in and see exactly where it is choking except for some cryptic, stupid error code. Not having a GUI also weeds out some of the people who don't have a clue from trying to use amanda :-) In the Windows department, I'm looking into using either built-in backup of XP Pro, or free software that came with my Snap for backing up Windows, and then use amanda to backup the Windows stuff off the Snap in the (hopefully rare) event of a 2-drive failure on the RAID-5. On Thu, 2004-04-08 at 15:35, Galen Johnson wrote: Does/can amanda utilize star? This would be one of the most useful capabilities I can foresee for Amanda (quiet Windows users). -- Jonathan Dill [EMAIL PROTECTED]
Re: Amanda backup strategy thoughts, full and incremental backups to tape once a week.
Here's a strategy that I implemented about a month ago that is working pretty well so far: 1. run amdump every night to large RAID w/o tape, mix of full and incr 2. run script to separate images to amanda-incr and amanda-full 3. when amanda-full exceeds size of tape, run amflush 4. when RAID approaches capacity, flush some incrs to tape Since it's a RAID, there is less chance it will go to crap than a regular disk partition, but it is still possible. And if it does, I still have the full dumps on tape to fall back on. To give you some gauge of capacities, I'm backing up about 300 GB of disks, have ~600 GB RAID5 (on a Snap 4500) and am using 100/200 GB LTO (Ultrium-1) tapes. So far, it's mostly web servers, but I haven't moved everything to the new backup cycle yet. I have flushed 2 tapes worth of full dumps, and accumulated about 167 GB of incrementals that are still on the RAID. The goal and idea is that 90% of the time, I will only need to load one tape to do a restore, and the rest will come out of holding files. Even better, there is a good chance that the tape that I need will be the last one that I flushed, so I won't have to change the tape. /snap/amanda-hd is the directory that amanda.conf looks for, which is actually a symbolic link to /snap/amanda-full. Therefore, I just move the incremental dumps from amanda-full to amanda-incr, both actually on the same filesystem (a simple directory rename in the filesystem inode table). If I need to flush incrementals, I just change the link amanda-hd - amanda-incr and run the flush (this could be done more elegantly with a patch, but this is so easy, so why bother?) In the cron job, I make sure amanda-hd - amanda-full before amdump runs. I'm using NFSv3 to XFS filesystem through a dedicated 100baseTX FD connection, so I'm using a chunksize of ~600 GB, and that makes the rename script very simple since I don't have to worry about any backup getting split into chunks. AFAIK NFSv3 and XFS both support 64-bit addressing, so I shouldn't run into any filesize limits. --jonathan
Re: Avg Dump Rate and Compression
There are a few more things to try. First, there may be an mt command to set the default compression for the drive--that will at least help make sure if you use some new tapes, they will get started with the correct compression. The tricky part is that some drives, such as 8mm, use the density to set the compression and ignore the compression command. You can usually get some ideas by looking at the output of mt status. Under Linux, the commands you want would be something like mt defcompress off and mt defdensity then the drive will (hopefully) use those settings by default when you load a new tape, and for some drives, may actually override what's on the tape that you load. I have sometimes put these commands into one of the system startup scripts, such as /etc/rc.d/rc.local on Linux. Second, some drives have dip switches that you can set to hard-disable compression. For example, I had a Sony DDS4 autoloader that had dip switches to do that. At least, the dip switches make it non-ambiguous what is going on. Third, some drives may have (usually Windows-based) firmware utilities that will let you set the settings of the drive, sort of like software dip switches. You would just need some Windows computer with a SCSI card in it, it wouldn't even have to be a good SCSI card just to send the commands to the drive as long as you have the right type of cable. The bad part is you have to disconnect the drive, possibly move it, and hook it up to the WIndows box. Worst case, maybe you could get some sort of SCSI adapter for a laptop with WIndows, such as PCMCIA, USB, Firewire, or even Paralllel to SCSI adapter. Fourth, some tape drives like AIT or AME (Mammoth) supposedly try to set the compression dynamically based on the data and don't care whatever you do to try to stop it--you would have to go to the manufacturer and look up the specs for the drive to find out if it does that. Lastly, when I have run into this problem, I usually have done this, but wonder if it could have caused problems: # load the tape mt compress off amrmtape config label amlabel config label If you're going to be doing this to a bunch of tapes in an autoloader, it probably would be easier to script the dd command that Gene described, because you just dump the header to a temp file and the script doesn't have to try to parse what's in the temp file to figure out what the tape should be called. Good luck! Gene Heskett wrote: The only way I know of to fix such a tape is: rewind it read out the label to a scratch file using dd rewind it again turn the compression off with mt -f device whatever turns it off dd that scratch file back to the tape using the non-rewinding device dd at least enough data from /dev/zero to force the drive to flush its buffers. This will force the drive to turn the header compression flag off and that tape will not be compressed again unless you turn compression back on. --jonathan
Re: selfcheck reply timed out
On Wed, 2004-03-24 at 11:30, Joshua Baker-LePain wrote: Remembet I have the NIS stopped. Then you need to either fix NIS or define the amanda user locally on the server. My experience has been that on linux, NIS (via ypbind) is unreliable and tends to crash, so I always put the amanda user in the local passwd file to make sure that backups don't fail even if ypbind craps out. I think part of my problem has been that I'm using SGI IRIX as NIS servers, and Linux as NIS clients, and something between them is not 100% compatible. If you really want to rely on NIS, I would suggest setting up some kind of a watchdog to restart ypbind if it fails. In fact, I think I am going to look into that option right now, it would fix some of the problems/complaints that I have had, like occasional problems with people not being able to login. -- Jonathan Dill [EMAIL PROTECTED]
Re: selfcheck reply timed out
Maybe there are some zombie amanda processes hanging in the background. Try looking at ps to see if any processes need to be killed. Look at the files in /tmp/amanda on the server and the clients to see if there are any clues. Try rebooting and see if the problem still persists. On Wed, 2004-03-24 at 13:09, Pablo Quinta Vidal wrote: I´ve NIS running without errors and now the message is selfcheck request timed out I had this problem earlier and when I solve this then came selfcheck reply timed out and now again the old problem. I checked the help ind docs/FAQ but it doesn´t work. Help please and sorry if I´m tedious!! Thanks -- Jonathan Dill [EMAIL PROTECTED]
Re: amanda and firewall
Stefan G. Weichinger wrote: on Dienstag, 23. März 2004 at 09:28 you wrote to amanda-users: PB (Don't be afraid, you wont hear any noise during compilation, it's PB quiet and easy :-) ) Seems to get the joke-of-the-week in here. Quiet funny ;) [sound of crickets chirping] I saw that in a comment on the blog Witt and Wisdom and just had to steal it :-) so please no offense. Joking aside, you could rpm -ivh the amanda src.rpm, and then hack the SPECS/amanda.spec file (usually in /usr/src/redhat or /usr/src/RPM) to change the ./configure command, then you would be able to rpm -bb SPECS/amanda.spec to compile a new binary RPM with the --portrange option that you wanted. --jonathan
Re: All estimates failed
Just a thought, but I would probably boot off a rescue disk and check the filesystem consistency on the client with the failing estimates. I'd probably use Knoppix just because it is user-friendly and provides a comfortable, gui environment for poking around. --jonathan James Tappin wrote: I've been having a similar problem in that as of 14 March all dump-based backups on my main machine have been failing because of an all estimate filed error. Tar-based backups are working correctly.
Re: flush full dumps only
I have attached a very simple but effective script that I set up to separate full and incremental holding disk files, so that full dumps can be flushed to tape while incrementals remain on disk. The script is almost so simple as to be a joke, but I don't have any idea what kind of bugs can be found. WARNING: The script is provided as-is with no warranty express or implied. If you screw up your amanda backups do not come crying to me. To really make the script work, you will need to change the values of $r and $i for your configuration, run it to see what it will do, and then remove the echo from the beginning of the two lines. If you can't figure out what that means, then you probably shouldn't be screwing around with this stuff anyway and should forget about it. For restores, I think I can flush the latest batch of full dumps, then use a modified amanda.conf that looks in the directory where the incrementals are stored, but I am not sure how amanda decides where holding disk files are located. I might have a use another script to move the incrementals back to the normal holding disk tree, but I will cross that bridge when I come to it, to use a rather hackneyed expression. -- Jonathan Dill [EMAIL PROTECTED] #!/bin/tcsh -f set r=(/snap/amanda-hd) set i=(/snap/amanda-incr) cd $r set dl=(`find . -depth -type d -print`) cd $i foreach d ($dl) if (! -d $d) then echo mkdir -p $d endif end cd $r foreach d ($dl) set fl=(`find . -type f \! -name *.0 -print`) foreach f ($fl) echo mv $f $i/$f end end
Re: High CPU running backup
I would guess that the ufsrestore is making an index of one of the dumps. If you don't care about interactive amrecover you could make a dumptype that doesn't do index that should eliminate the ufsrestore process. Running fewer dumps in parallel should help, too. I don't know a lot about Solaris. 0.0% swap should mean nothing is paging, but with 729M swap used, I'd still check out high-memory processes, maybe something is leaking memory, but any paging should still show up in swap and not kernel I would think. It looks to me like it wouldn't take a lot to make the system start paging heavily from the state that it is in with only 13M of real memory free. Simon Lorenz wrote: Any suggestions for stopping this would be much appreciated. load averages: 2.21, 2.21, 2.02 17:48:54 142 processes: 131 sleeping, 5 running, 1 zombie, 4 stopped, 1 on cpu CPU states: 0.0% idle, 13.7% user, 76.9% kernel, 9.3% iowait, 0.0% swap Memory: 768M real, 13M free, 729M swap in use, 3911M swap free PID USERNAME THR PRI NICE SIZE RES STATETIMECPU COMMAND 10563 amanda 1 220 2232K 936K sleep5:32 35.03% sendbackup 10476 amanda 1 0 19 2640K 1920K run 2:26 15.04% dumper 10568 amanda 1 320 11M 9928K sleep1:49 12.55% ufsrestore 10572 amanda 1 480 11M 2864K run 0:47 5.20% ufsdump 10571 amanda 1 530 11M 2864K run 0:48 5.16% ufsdump 10573 amanda 1 480 11M 2872K run 0:47 4.64% ufsdump 10574 amanda 1 360 11M 3072K sleep0:32 3.53% ufsdump 10570 amanda 1 480 11M 7520K sleep0:20 1.89% ufsdump 10646 root 1 580 2104K 1208K cpu 0:01 0.35% top
Re: I don't think Amanda is going to work for my environment...
On Thu, 2004-03-18 at 14:31, Joshua Baker-LePain wrote: It is true that it takes some wedging to make amanda work in a 'doze heavy environment -- that's simply not what it was designed for. As for advice on commercial solutions, this isn't exactly the best place to ask. ;) If you're not on a shoestring budget, I'd say knock yourself out, there are some great (but usually expensive) packages out there that are very easy to set up and maintain, though this list is not the best place to get that sort of information. Veritas, Arkeia, and Legato are three packages I would definitely check out if money was not an object. You could, however, consider a sort of hybrid approach to handle the 'doze machines, which is what I am thinking about right now. In a nutshell, you would use some other mechanism to make the 'doze machines back themselves up to a share, and then you would use amanda to backup that share periodically. Windows XP Professional has a built-in backup/restore function, which is one option that I plan to try. I am using a 1 TB Snap Appliance 4500 NAS which came bundled with Backup Express for Guardian OS and an unlimited number of Windows clients, so that is another option that I plan to try. The Snap itself runs Guardian OS, which is some flavor of Linux, and I have it NFS mounted on my amanda server, and also use it as my holding disk for amanda backups. Then I can use GNUTAR with amanda to backup the directory where the 'doze machines have taken a dump. A lot of Windows people I have talked to lately are using a 2nd hard drive on USB or Firewire to do backups, or in a cold-swap cartridge. Norton Ghost is a popular application to do that, also XP Pro backup/restore, and even robocopy (is that Win2K Pro?) with varying degrees of automation. DVD+R with 4.7 GB capacity is popular for longer-term storage. The latest Norton Suite 2004 Pro looks like makes it pretty easy to automate Ghost backup to a 2nd disk or to DVD+R (I have it on the Windows partition of my home computer) but I mainly use Windows for games and don't care enough about losing that data to have put the effort into automating backups. --jonathan
flush full dumps only?
Has anyone tried the approach of only flushing full dumps and leaving incremental dumps on disk? I think that this would have roughly the same effect as doing full dumps out of cycle, but I have not had luck so far with the out-of-cycle approach. I think that it should be fairly easy to script, create a parallel directory tree and temporarily move out the incremental dumps that you don't want to flush, for example. The idea would be to leave all of the incremental dumps available on a huge disk, but keep all of the full dumps on tape, so at most one tape should be required for any restore. Since the full dumps take up the most space, clearing them out should leave more space for a longer history of incrementals. At least that is the idea. --jonathan
Re: filesystem limit?
Geoff Swavley wrote: I was wondering if anyone could tell me why amanda seems to split my filesystem into 2 holding files? An error has occurred so these filesare It sounds like you have other, unrelated problems, but check the setting of chunksize in amanda.conf, that is usually what determines dumps getting split up into multiple holding files. In any case, it is not determined by the filesystem--you can give amanda a chunksize that is too large for some filesystem type, in which case it will keep writing until it gets an I/O error. The holding files are concatenated during writing to tape, at which point the holding file size becomes irrelevant. Most modern filesystems can support file sizes up to the TB range--I'm using a chunksize of 256GB on Linux XFS, it could be larger but Why go larger than the size of the holding disk? Exceptions include (but are not limited to) ext2, SGI efs, iso9660, FAT(12|16|32), NFSv2, not sure about ufs, hfs, NTFS, I'm sure there are others. 2 GB and 8 GB are typical limits. But as I said AFAIK amanda will not automagically set the chunksize for the holding disk filesystem type, it would just keep writing until it got an I/O error, but then there have been a lot of changes from 2.4.2 to 2.4.4 that I am just catching up with now. --jonathan
Re: Disaster Recovery
Joshua Baker-LePain wrote: I think what you mean is what files do you need in order to save the complete current state and history of the backups, although I'm guessing as your request was overly terse. If that's right, you need: the config dirs (where your amanda.confs are) the infofile dirs as defined in your amanda.confs the logdirs as defined in your amanda.confs the indexdirs as defined in your amanda.confs How I deal with it is that I just rsync the amanda account home directory to another server periodically. If you wanted to, you could probably set it up as a daily cron job or something. --jonathan
Re: *Slow* amrestore
Joshua Baker-LePain wrote: I'm reading some somewhat large (14-18GB) images off of AIT3 tapes, and it's taking *forever*. Some crude calculations show it coming off the tape at around 80 KB/s, whereas it was written out at 11701.6 KB/s. The tapes were written in variable block size mode. What's the best way to read these images more quickly? Are you piping the amrestore output, say to uncompress and extract files? Maybe the extraction process is too slow, and causes the tape to have to stop and reposition while the pipe is emptying out. I wonder if you could put a bigger buffer in the pipe somehow, that might help, say 64 MB to begin--I remember seeing a cat-like program that could buffer a pipe, I think it was with Linux software for doing backups to CD-R. I would try amrestore -c to just dump the image off the tape, and then do the uncompress and extraction separately, but you will need enough disk space to do it. Worst case, you could try amrestore -r and use dd bs=32k skip=1 if=dump-file to skip over the header, and then pipe that to the uncompress and extract processes. --jonathan
Re: *Slow* amrestore
Hmm. Check also mt status to make sure the drive thinks that the blocksize is 0 if not change it with mt blksize. The files will be perfectly fine with bs=16M. gzip and/or tar will probably give a warning bitching about the nulls at the end, but it won't have any effect on the restore, the files will still be intact. In some desparate cases in the past, where some mung brain had data on Sony Video 8 tapes (!) I ended up mucking around with mt blksize and just using cat to try to grab the stuff off the tape without any blocking, crazy as that sounds, but that was also on SGI IRIX with Exabyte 820 Eagle drives. It was a pain, and there were loads of errors, but I got most of the data off the tapes. --jonathan Joshua Baker-LePain wrote: On Thu, 11 Mar 2004 at 2:21pm, Jonathan Dill wrote I would try amrestore -c to just dump the image off the tape, and then do the uncompress and extraction separately, but you will need enough disk space to do it. Worst case, you could try amrestore -r and use dd bs=32k skip=1 if=dump-file to skip over the header, and then pipe that to the uncompress and extract processes. The current slow-running process is 'amrestore -r'. I'm guessing it's forcing the block size that's the culprit here -- the amrestore man page isn't quite clear, but it seems to say that if you don't specify -b, it'll default to 32k. I tried just 'dd if=/dev/nst1 of=img.raw' but it said dd: reading `/dev/nst1': Cannot allocate memory. tapeinfo says this about blocksizes for the drive: MinBlock:2 MaxBlock:16777215 Maybe I should try dd with bs=16M? But will that pad the output file with an unacceptable-to-tar chunk at the end since the tapefile is unlikely to be an exact multiple of 16M? Thanks.
Re: *Slow* amrestore
Have you taken a look around in /proc/scsi? /proc/scsi/scsi should give you some basic information, and the subdir for your driver should give more details, such as what transfer rate the drive is negotatiated at, for example /proc/scsi/aic7xxx/0 for an Adaptec 2940 series. Perhaps there was a bus problem and the device hooked up at 10 MHz async or something. Reboot and get into the SCSI bios config page and make sure everything in there is kosher. I have even seen a few terminators go bad, so such things are not impossible, even though the drive was writing fine before. Since dd is having just as much problems, I would suspect that amanda is not the problem per se. Could be kernel, could be hardware, could be the tape cartridge, could be some stupid, proprietary Sony thing, dip switch settings on the drive, hardware compression settings. In some cases, I have taken tape drives and hooked them up to a Windows box so I could run proprietary diagnostics, take a look at the tape drive's internal logs, update the drive firmware. I think the weirdest stuff was using a 9600 bps null modem serial terminal connection to connect to the embedded console on an Exabyte 10h library, very strange set up. I have also pulled out SCSI cards and put them on a Windows box to check out the SCSI BIOS and load new firmware. Run a cleaning tape through just in case--with some drives, I had the problem of drives actually needing cleaning more frequently than the drives thought they needed it. Exabyte Eliant drives were notorious for that--I would run into these problems where I had to run a cleaning tape through 5-6 times in a row (even though the drive said it didn't need cleaning) and then finally it was fine, but in those cases, I was also seeing read and/or write errors. The Eliants just based cleaning on hours of read/write time, which turned out not to be often enough, so crud would accumulate over time--no doubt because amanda was streaming so efficiently that the drive was actually processing more length of tape than ordinarily anticipated :-) --jonathan
client with private address
I want to backup a client on a private network 10.160.32, but amanda seems to be looking for a DNS to resolve the IP, and then do a reverse lookup on the IP to get the hostname. Is there a way to do this without setting up a DNS for 10.160.32? I wish amanda would just believe the address that I put in disklist instead of double-checking with a DNS. Does she not trust me? Am I not trustworthy? I guess I'll pick at the docs for gethostbyaddr and gethostbyname calls to see if there is any way to modify their behavior, or if there is a different routine that I could plug in that would check /etc/hosts first, or something like that. --jonathan
Re: client with private address
Nevermind--I just added the address of the server to /etc/hosts on the client and that fixed the problem. Some very useful information in man gethostbyname and man gethostbyaddr /etc/host.conf may be consulted for the res order for those two calls. The default is to check bind first, but you can override that by putting order hosts, bind in /etc/host.conf. At least that's how it works on some flavors of Linux. Jonathan Dill wrote: I want to backup a client on a private network 10.160.32, but amanda seems to be looking for a DNS to resolve the IP, and then do a reverse lookup on the IP to get the hostname. Is there a way to do this without setting up a DNS for 10.160.32? I wish amanda would just believe the address that I put in disklist instead of double-checking with a DNS. Does she not trust me? Am I not trustworthy? I guess I'll pick at the docs for gethostbyaddr and gethostbyname calls to see if there is any way to modify their behavior, or if there is a different routine that I could plug in that would check /etc/hosts first, or something like that. --jonathan
Re: client with private address
A common shorthand for specifying a Class C subnet is to leave off the 4th number, basically the same thing as 10.160.32.0, 10.160.32.0/24, or 10.160.32.0/255.255.255.0 etc. On Tue, 2004-03-09 at 15:26, Hans-Christian Armingeon wrote: Am Dienstag, 9. März 2004 20:11 schrieben Sie: I want to backup a client on a private network 10.160.32, but amanda I think, that an ipv4 address has four address parts, you have only three. -- Jonathan Dill [EMAIL PROTECTED] jfdill.com
Re: mixed full / incremental backup...
Here's a very simple solution: 1. reserve 100 in amanda.conf (or comment out reserve line) 2. leave out the tape during the week (*or change dev to /no/such/tape) 3. run amflush before Friday backups In this case, amanda should try to do degraded mode backups during the week while there is no tape, only incrementals. On Friday, you put a tape in the drive before the backups, and I think it should try to catch up on some full backups. It probably won't give you perfect incr/full separation, but it would be good enough for me. Naturally, you will need sufficient holding disk space. You will have to keep an eye out for messages warning about overwrite last full dump. * do not change tapedev to /dev/null because that activates a test mode where amanda will happily dump all data into the bit bucket rather than decide there is no tape. --jonathan Jon LaBadie wrote: What about a single config. Set it up with a long dumpcycle. Use dumptypes that specify always incremental, or no-full, or strategy incr or ??? I.e. something that reduces the possibility of full dumps.
Re: client with private address
Hi Frank, The documentation for gethostbyaddr and gethostbyname explained how each call goes about looking up addresses. At least under Linux, there were several opportunities to override the default behavior and make the routines consult /etc/hosts first. In my particular case, there are only two private addresses that I need to handle due to the amanda server and client having a direct cross-over connection, for an unrelated purpose. For two IP addresses, it really didn't seem worth it to set up a local DNS with forward and reverse domains. As for address spoofing, there are basically 2 scenarios that I can think of: 1. idiot hacker causes some backup(s) to fail on one night, maybe a DoS, but that's about the extent of it 2. hacker who knows about amanda, and has the right ports open to intercept and capture the stream, possibly to steal sensitive data #2 would probably be loads easier to do with just a run of the mill sniffer that can capture streams, and the activity would be much less likely to be detected. I can't see the benefit of impersonating the amanda server, besides which it would cause loads of errors and send up red flags that something was going on. Not to mention that if your data is all that sensitive, you should really be encrypting the data on the client and not sending it in the clear across the network, and the systems should be behind a tight firewall if not disconnected from the internet altogether. I really can't imagine DNS spoofing being that big of a risk with respect to amanda. Having the addresses hard coded in /etc/hosts and looking at that and not the DNS should be more secure than relying on DNS lookups crossing the network, which could be spoofed. Frank Smith wrote: I suspect that Amanda was designed to use hostnames in their disklists and .amandahosts, and names are very easy to spoof, so the lookups are done to verify that the correct host is connecting. I'm sure the code could be modified to not do lookups if given an IP, but having proper DNS has many other benefits than just helping Amanda.
Re: Strange DNS lookup problems ... I think ...
I read the last e-mail about this, but lost it, but I think I remember the basic details. First, I would try setting up some sort of nameservice caching on the client and server as a work-around. Some flavors of Linux have a caching-nameserver package that sets up the correct bind files for you, then you just put nameserver 127.0.01 at the top of /etc/resolv.conf. tmdns is supposed to be a more lightweight caching nameserver of some sort, but I haven't had good luck with it so far. nscd is a more general-purpose nameservice caching mechanism that can also cache NIS and LDAP data, but I think there may be a kernel piece to it that you also need compiled into the kernel. SGI IRIX has nsd which is similar to nscd. If you use nscd or nsd, check /etc/nsswitch.conf for the order that name services will be checked for hosts. In particular, you may need to delete nis+ or nisplus if you don't have NIS+ running on your network--It is often in there as one of the defaults, but can cause the host res process to crap out at that point if you don't have NIS+ available on your network. Second, I would check interface statistics on the client, server, nameserver, and switches and routers if possible. You want to check for collisions and/or errors, and keep an eye out for duplex mismatch or auto-negotiation problems related to certain hardware. Watch out for misbehaving mini-hubs or mini-switches along the way. I have had problems with interface hotplug on Linux and certain cards not detecting a link or auto-negotiating correctly, eg. 3c509B. I had to put MII_NOT_SUPPORTED=yes in /etc/sysconfig/network-scripts/ifcfg-ethX where X is the number of the interface, to explicitly disable hotplug for that adapter. --jonathan
hybrid theory
Is anyone successfully using a mixed strategy of backups to both disk and tape? In particular, I have a 1 TB Snap 4500 and an LTO tape drive. I have thought about a few different ways to go about it, but would appreciate suggestions from someone who has tried this approach. The Snap also has built-in snapshot capability, and I wondered if there was a way to make use of that in combination with backups to disk. 1. normal amanda backup to holding disk, flush when disk is getting full 2. do full dumps to tape out of cycle and incrementals to disk with file-driver 3. #2 but also use snapshot technology to keep an even longer history of incrementals 4. use file-driver, and occasionally archive some of the tape-files to real tape Any other ideas? Thanks, --jonathan
Re: Strange DNS lookup problems ... I think ...
Resolving IP address to a hostname (reverse lookup) is the part that looks broken, check the reverse domain in the DNS i.e. host 172.24.16.86 or nslookup 172.24.16.86 The error says *hostname* lookup failed, not address lookup failed. Someone else reported a similar problem a few days ago, and he reported that there was a typo in the reverse domain file of the DNS, and that fixing that fixed the problem. I wonder though, Why does amanda need to do a reverse lookup? You give amanda a hostname in the DLE, and it looks up the IP address, which should be adequate for amanda to do what it needs to do. But then it tries to do a reverse lookup for the hostname based on the IP address, and gives up if that fails. It would be nice if the reverse lookup could be avoided. In principle, yes, the reverse table in your DNS should be correct, but failing backup seems like an expensive DNS diagnostic. planner: ERROR intrap: [addr 172.24.16.86: hostname lookup failed] --jonathan
Re: server lookup failing
I have run into this problem before with NIS when ypbind crashed on some of the clients--this has been a chronic problem for me with Linux talking to IRIX NIS servers. Consequently, I put the amanda host IP, amanda user and group IDs in the local files so that ypbind crashing would not muck up the backups. amanda doesn't cache any IPs AFAIK, but the OSes of the clients probably do. Also, special case if you are using NIS--you have to make sure hosts is updated in NIS and you have re-run ypmake manually, plus in IRIX 6.X you may need to do the nsadmin commands on the NIS master. Also, check the clients to see what's in /etc/nsswitch.conf if it exists--that should tell you what data sources are checked and in what order. It's interesting that host works and amanda does not--perhaps they use different host lookup mechanisms, and you could get clues from looking at the source code of both programs. As for the clients, it's going to vary depending on the client OS, but here are a few I can think of off the top of my head: # IRIX 6.X nsadmin flush nsadmin restart # Linux with cacheing-nameserver /etc/rc.d/init.d/named restart # Linux with tmdns /etc/rc.d/init.d/tmdns restart George Kelbley wrote: After changing the IP on my amanda server, many of my clients fail amcheck with host name lookup fail. The server is resolved via dns (not in /etc/hosts), running host on the clients returns the proper IP. Re-booting the client fixes the problem, but I'd rather not reboot all my clients. So, does amanda cache the server IP somewhere that I can flush? I been looking around but have't found anything. thanks
Re: gnutar in configure
If you're backing up more than one architecture, I find that it's nice to set things up so that you can have the same path to gnutar on all of the architectures. That way, you can run amrecover on any machine and it will find a valid path to gnutar. Normally, I just create a symbolic link to the real path to gnutar in /usr/local/etc/amanda/bin and use --with-gnutar=/usr/local/etc/amanda/bin/tar If you're only backing up one architecture, and this suggestion causes a brain hemmorhage, then forget about it and just stick to what you were doing. Maybe it seems a bit esoteric, but in practice I have found it to be very useful. Very often, I have restored files for an IRIX workstation to the holding disk of the amanda server, which is Linux, and then pick and choose which files I really want to transfer to the workstation with rsync, rather than just restoring everything to the IRIX workstation directly. That way, I can be careful not to overwrite files, or force overwriting corrupt files with newer timestamps, whatever it is that I need to do. I think it gives me an extra level of control to help avoid making a mistake. On another note, maybe things have changed, but I once found that gnutar incremental backups sucked performance-wise, would make machines pretty much unusable during estimates and dumps. Normally, this would not matter, but you're talking University with eccentric grad students working at 3am and such who complain about these things. I have migrated most things to XFS filesystem and use xfsdump on Linux and IRIX--a process that I started when XFS went Open Source (around Red Hat 7.0) and I got tired of waiting for the problems with dump for ext2fs to get sorted out. Machines are still very usable with xfsdump and software compression running in the background, and finish faster than gnutar dumps. xfsdump estimates are very fast, comparatively speaking. However, with faster CPUs, faster disk interfaces, and filesystems like Rieserfs, perhaps the performance of gnutar has improved. --jonathan Frank Smith wrote: Then you can run configure --with-gnutar=/usr/local/bin/tar, and make sure that that path exists on your clients, and is gnu tar of the proper version on all of them as well.
fake install path
I vaguely recall that there is variable that you can pass to 'make' to install to a different root, similar to what happens during building a binary RPM, for example: make install VARIABLE=/tmp/amanda-2.4.4p2 The result is that you end up with all of the amanda files in /tmp/amanda-2.4.4p2/usr/local and so on. Then you can do something like this: cd /tmp/amanda-2.4.4p2 tar cvzf amanda-tarball.tgz . Then you can ftp the tarball to a different machine, cd / and extract the tarball. Does anybody remember what VARIABLE is supposed to be or how to do this? My goal is to set up a Snap Appliance 4500 as an amanda server--Yeah, I have already seen Preston Smith's page about setting up amanda client on a Snap 4400, and I also know that Snap Appliance won't take responsibility if I somehow manage to screw up my Snap. Thanks, --jonathan
Re: fake install path
Thanks, that worked: make install prefix=/tmp/amanda-2.4.4p2/usr/local cd /tmp/amanda-2.4.4p2 tar cvzf ../amanda-tarball.tgz Paul Bijnens wrote: The prefix= can be specified to override the configure prefix like: make install prefix=/tmp/amanda-2.4.4p2
Re: dds3 or dds4?
Tom Strickland wrote: We're on the verge of ordering a DDS drive (this week). It'll probably be an HP Surestore - but the question is DDS3 or DDS4? There's the obvious difference in capacity, but beyond that are there any other differences? Speed is an obvious one - any others? Keep in mind that amanda can't span individual backups across multiple tapes and think about your requirements. If you buy a new disk today, 18 GB is probably the smallest size that you will find easily available, and that number keeps going up. I work in an academic research lab where now some people are getting 75 GB and 180 GB disk drives, and with new instruments the data sets are increasing in size rapidly. Even the DDS4 isn't big enough anymore, so I'm having to look into getting a tapedrive with higher capacity (or using GNUTAR to split up dumps, or partitioning the drives into smaller chunks, both of which are kind of messy solutions). If things are different in whatever business you are in, maybe the DDS3 would be adequate, but it's something to think about. -- Jonathan F. Dill ([EMAIL PROTECTED])
Re: dds3 or dds4?
Hi Tom, OK I got it now this is a charity. If the backups aren't too big and you have a big enough holding disk, you might consider a strategy where you do some dumps to the holding disk. For example, I have a config where on Wednesdays, I flush the holding disk and then do a dump with the tape in the drive, and the rest of the week I just dump to disk. If you are just backing up one or two servers, you might consider using a tool like rsync to mirror the disks to another server off site that is not web accessible, or ghosting to a removable hard drive. Lastly, you might consider looking for a corporate sponsor who could put the $ for the DDS4. Tom Strickland wrote: I've just done some sums: total for HP SureStore DDS3i, SCSI card, 20 tapes, delivery, VAT: 998.52UKP total for HP SureStore DDS4i etc: 1408.71 UKP total for HP DLT1, SCSI card, 20 tapes, delivery, VAT: 2441.415 Well, unless I'm being wildly ripped off somewhere, it looks as though DDS is the only affordable solution. Probably DDS3, I'm afraid. I may be able to work something out, but at the moment I don't have the funds to be able to chip in myself. Anyway, thanks to all for the advice. Very helpful. -- Jonathan F. Dill ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill
Re: Try2, Amanda Question
Bryan S. Sampsel wrote: That's the bitch of it...it IS resolved: via nslookup, via ping--ANYTHING but Amanda. It's bizarre. I'm getting ready to compile amanda from source to see if it's a problem with the rpm on the client. Rpm installs are OK sometimes--other times, I'd rather not deal with 'em. Do you have all of the redhat updates applied on the client? nscd was buggy in RH 7.0 and that could cause problems with hostname lookups if you use nscd and have not upgraded to the updated rpm. There were also updated rpms for bind-utils, glibc, kernel, and possibly some other packages that could affect hostname lookups. You might want to run nscd on the client to do nameservice cacheing. Check out /etc/nsswitch.conf and possibly put files first for hosts, but then be careful about keeping /etc/hosts up to date or only put the minimum necessary entries in /etc/hosts. If you use nscd, you might have to do the following to flush the cache (on the amanda client, because that is where the lookups are failing): /etc/rc.d/init.d/nscd restart Stupid question, but did you add the address to /etc/hosts on both the server and the client? -- Jonathan F. Dill ([EMAIL PROTECTED])
Re: Raid and Amanda
Olivier Nicole wrote: I am learning to use Amanda but it seems it has a problem, as everyone knows, backing up a filesystem larger than the tape. Am I wrong? I understood that Amanda is supposed to ask for as many tapes during a single run, that are needed to complete the back-up. So it would start and ask for a second, third, fourth... tape until it managed to back-up everything. A backup of an individual filesystem cannot span multiple tapes. For example, if you have a 40 GB tapedrive, and you want to back up a 75 GB disk drive that is pretty full of data that is not very compressible, it's probably not going to work. You will have to use gnutar to split up the disk into smaller subdirectories. Or, you might consider backing up the disk and then repartitioning it into 40 GB chunks. I think that spanning multiple tapes may be a feature in some development version of amanda, but I don't know the current status. Secondly, runtapes determines the total number of tapes that amanda will use during the run, period. You could set runtapes equal to the number of tapes in your changer and let amanda use all of the tapes until it is done, but that may not be the ideal strategy in the long term. You may want to try to get amanda to use about 2-3 tapes a day, for example, and balance the full and incremental dumps accordingly. You have to try to be realistic about how many tapes you will need to use, but things will also balance out some as the dump cycle progresses. What I do is during the week, I only let amanda use 2-3 tapes for each run. Sometimes, dumps end up left on the holding disk, and I have to flush them to tape, but at least it keeps the backup cycle from running into the middle of the work day if there are some large dumps that need to get done. For the weekend, I let amanda use all of the tapes in the rack if it wants to. I will also deliberately force a few full dumps to try to get them done on the weekend so they won't run during the week. One thing I haven't worked out yet is a good system for trying to guess what full dumps should be coming due soon so I can specifically force those dumps on the weekend. -- Jonathan F. Dill ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill
Re: Is amanda suitable for a single server backup?
Hello Howard, There are two solutions that I have used in this situation: 1) If you can handle a few minutes of down time, like on a home PC or personal workstation, you could use ghosting to a removable disk drive. This is probably OK for a home PC or another situation where you can handle losing some data, because if both disks fail, you would lose everything, so that is why it's a good idea to remove the backup disk and store it offline. Or, if you are on a network, the backup disk could be on a another computer. (NB: In case you were thinking of it, RAID1 or simple mirroring is NOT a backup solution and will not cover you for eg. if a hacker trashes the filesystem). 2) You could use amanda to do backups to holding disk most of the time, and flush the dumps to tape when the holding disk is getting full. Since you have 40/80 tapes, if you use software data compression, you would want about 40 GB holding disk to get the most out of the tape. If you can use an IDE disk like ATA-100, in the US you can get a 40 GB disk for about $120-150 USD (see www.megahaus.com for example). If you look for a close-out you can probably get it even cheaper, but may have problems getting a warranty replacement if it fails. If both disks fail, you will lose everything up to the last time that you flushed the holding disk to tape. For this reason, it would be better to use another computer as the backup server if possible. Howard Zhou wrote: If I have a single server to backup with 4Gb of data backed up to a 40-80Gb tape without a tape changer, would amanda do the job? To my understanding, amanda does append to a tape. Without appending, you have to put in a new tape everyday for a small amount of incremental backup data on 40-80Gb tape (what a waste) plus manual process. Is there any plan to support appending to tape in a future release? Is there any other free backup system which supports this feature? Thanks in advance. Howard -- Jonathan F. Dill ([EMAIL PROTECTED])
predicting full dumps
Does anyone have any scripts, or any tricks that I could use to try to predict what full dumps are coming due? I would like to be able to force a bunch of full dumps on the weekend, when the total time of the run is not an issue, so that the load should be a little lighter during the week. Thanks in advance, -- Jonathan F. Dill ([EMAIL PROTECTED])
Re: Fwd: Re: predicting full dumps
Hi guys, Actually, as Joshua pointed out, amadmin config due does exactly what I want--It gives you a nice list saying which dumps are overdue, what dumps are due today, and how many days until other dumps are due. Forcing dumps that are overdue, due today, or that are due in a day or two would be a good starting place. You could do something like this to consider candidates for forcing: amadmin config due | fgrep -v Overdue | fgrep -v today \ | awk '{print $3,$0}' | sort -rn amadmin config due | fgrep today amadmin config due | fgrep Overdue You would start from the bottom of the list and work your way up. Mitch Collinsworth wrote: You can figure out what's coming due by looking at 'amadmin config balance'. You can't always predict what might get promoted by planner, but if you do what you're suggesting and force enough filesystems to make the total data size for the run greater than average, planner won't promote anything. Some other things to consider if run-time is getting to be a problem are adding more holding disk space and possibly increasing maxdumps. These should allow for more parallelism in the dumping phase. Depending on your needs, getting the data off the clients quickly might be the real issue, and then amanda can continue spooling data from the holding disk to tape long after the dumps have finished. -Mitch On Thu, 7 Jun 2001, Francis 'Dexter' Gois wrote: Not sure it is possible : the dumps queue is created when you launch amdump, and Amanda looks at which dumps it can move (to optimize) or it has to move (because of space problems) just before starting the dumps when you launch amdump Config (see the logs). Dexter On Thursday 07 June 2001 06:48, you wrote: Does anyone have any scripts, or any tricks that I could use to try to predict what full dumps are coming due? I would like to be able to force a bunch of full dumps on the weekend, when the total time of the run is not an issue, so that the load should be a little lighter during the week. Thanks in advance, -- Knowledge-sharing and open-source content : another way to gain eternity. Francis 'Dexter' Gois - [EMAIL PROTECTED] System Administrator - Tiscali Belgium NV/SA phone: +3224000839- fax : +3224000899 --- -- Knowledge-sharing and open-source content : another way to gain eternity. Francis 'Dexter' Gois - [EMAIL PROTECTED] System Administrator - Tiscali Belgium NV/SA phone: +3224000839- fax : +3224000899 -- Jonathan F. Dill ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill
Re: Linux and dump
Hi Eric, You may want to take a look through the list archives at: http://groups.yahoo.com/group/amanda-users/ This subject has already been hashed and rehashed to death on just about every mailing list that I subscribe to including this one. I'm planning to migrate to SGI XFS on Linux--SGI has released an installer CD for Red Hat 7.1 which can make XFS filesystems. XFS is a journaled filesystem, and it can be run over RAID unlike ext3 which had problems with RAID on 2.2 kernel. You can download the installer for free from ftp://linux-xfs.sgi.com but the server appears to be down right now. Eric Veldhuyzen wrote: I just saw that someone had problems with dump and Linux. This made me remember an posting from Linux Torvalds of a few weeks back which I think anyone still using dump with Linux should read: http://www.lwn.net/2001/0503/a/lt-dump.php3 Summary: Dump was a stupid program in the first place. Leave it behind. -- Jonathan F. Dill ([EMAIL PROTECTED])
Re: Linux and dump
I think I was the one who made the suggestion to remove ext2 dump, but I was wrong, you don't have to do that. ./configure will find both xfsdump and dump, and amanda will choose whichever program is appropriate for the type of filesystem i.e. if it is XFS filesystem and you have not specified GNUTAR, xfsdump will be used automatically. Alexandre Oliva wrote: On May 17, 2001, C. Chan [EMAIL PROTECTED] wrote: I followed suggestions to remove ext2 dump so Amanda would detect xfsdump and recompiled, but I find this rather inelegant. What is inelegant? Removing ext2 dump? You didn't have to do that. You only need xfsdump available at configure time to get the xfsdump supporting bits enabled; the existence of ext2 dump doesn't make a difference. -- Jonathan F. Dill ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill
Re: amanda
You can read the tapes without amanda using just dd and a restore program. You can get some hints by looking at the first part of a dump image. For example I will do these commands with one of my backup tapes. First, I have to mt fsf to skip over the tape header, and then I can use just dd command to read the image header which has the instructions: [root@masq1 ~]# mt fsf [root@masq1 ~]# dd if=/dev/nht0 bs=32k count=1 AMANDA: FILE 20010503 shaman /mnt/raw lev 1 comp .gz program /usr/local/etc/amanda/bin/tar To restore, position tape at start of file and run: dd if=tape bs=32k skip=1 | /bin/gzip -dc | usr/local/etc/amanda/bin/tar -f... - Maybe your backups were made with ufsdump or dump or some other program than tar, but the dump header should tell you what to do. Also, if you go to www.amanda.org and find your way to the amanda FAQ I think the instructions are there, or if you download the source code and unpack it, I think you can find them in some files in the docs directory. Also, don't forget to always make sure that your tape is write-protected when you are dealing with old backup tapes. I think this is just some good practical advice for any kind of backup, and you could have big trouble if you forget and someone else decides to use the tapedrive. PQ20 Yuval Ruttkai wrote: I am working in a company that their first backup was made before 4 years by Amanda. There are old tapes that they want to see if they need the material. I need to install amanda just to watch the files that are in the tape. I have Sun Ultra 5 with Solaris 5.8. Do I need a special configuration to configure Amanda just to watch and read those tapes? Please try to help me as soon as you can. -- Jonathan F. Dill ([EMAIL PROTECTED])
Re: file system problem - umount
Marcelo G. Narciso wrote: | DUMP: Warning - cannot read sector 600384 of `/dev/rdsk/c0t6d0s0' | DUMP: Warning - cannot read sector 600400 of `/dev/rdsk/c0t6d0s0' It looks to me like you have some bad sectors on your disk, or possibly a disk drive that is on its way to failing, like the head is having trouble seeking to some sectors. The filesystem was probably automatically unmounted to prevent corruption, or at least I have seen something similar happen with an XFS filesystem on SGI IRIX. Check the system logs. After amdump, the file system /dev/rdsk/c0t6d0s0 is unmounted. Why it happens? Someone knows what is happens? Thanks a lot. -- Jonathan F. Dill ([EMAIL PROTECTED])
Re: Dumping live filesystems on Linux 2.4 - Corruption
C. Chan wrote: I'm experimenting with XFS now and if I run into any glitches I'll let the Amanda list know. I'd like know how to tell Amanda to use xfsdump rather than GNU tar on XFS partitions. I think you would have to recompile amanda. I would install xfsdump/xfsrestore and rpm -e dump and then run ./configure script, then it should find xfsdump and choose that as the OS dump program. Then you would assign a non-GNUTAR dumptype to the disk(s) and it should run xfsdump for the backups. Hopefully, the configure script does not check to see if the OS is IRIX or not Linux and decide that xfsdump can't be installed on the system--Then you would have to hack the configure script a little bit to get it to work properly. -- Jonathan F. Dill ([EMAIL PROTECTED])
Re: Linux kernel 2.4 and dump vs. SGI XFS
Hi folks, If you are concerned about dumping live filesystem with Linux kernel 2.4, you might be interested in this: SGI has released an updated anaconda installer CD for Red Hat 7.1 with XFS journaled filesystem support. You load the installer CD first, which includes only the RPMs necessary for XFS support, and then you will need the regular Red Hat 7.1 CDs to complete the installation: ftp://linux-xfs.sgi.com/projects/xfs/download/ With XFS you have the option of using xfsdump for your backups on Linux--AFAIK this should work with amanda, but I have not tested it and YMMV. According to my discussions with Eric Sandeen at SGI, xfsdump does not look at the on-disk metadata structures directly, so it would not be subject to the problem which affects ext2fs and Linux dump. I am forwarding Eric's response to a couple questions that I had about it. Now if only amrecover would work with xfsdump, I would be in heaven :-) Original Message Subject: Re: SGI's XFS: ready for production use? Date: Fri, 11 May 2001 11:39:19 -0500 From: Eric Sandeen [EMAIL PROTECTED] To: Jonathan Dill [EMAIL PROTECTED] CC: [EMAIL PROTECTED] [EMAIL PROTECTED] References: [EMAIL PROTECTED] Jonathan Dill wrote: Hi Eric, Are there plans to keep up with new releases of Red Hat eg. release a new anaconda disk as new versions come out? I have a mostly SGI IRIX shop with a growing Linux population, so it seems somehow attractive to be able to use XFS as a standard--It could mean good things for consistent, automated backups, and being able to toss filesystems around between IRIX and Linux. I would consider adopting this as a standard, but I want to make sure that it will be updated and not a dead end. In terms of Red Hat integration, we don't really have an official guarantee that we'll keep up with the Red Hat releases. However, I expect that we will, unofficially if not officially. There are enough people interested in this inside and outside of SGI that I think this will continue. I have a personal interest in continuing. Of course, if Red Hat and/or Linus picks up XFS, that would be a good thing too, and we're trying to facilitate both of those things. As far as SGI's commitment to XFS on Linux, while I can't speak for the company, I see no signs of that commitment waning. We need XFS on Linux for upcoming products, and for our customers. And of course, the source is out there, so that's a certain level of guarantee that XFS will be around. See also the Long Term Commitment... thread at http://linux-xfs.sgi.com/projects/xfs/mail_archive/0105/threads.html Also, there has been a lot of chatter lately about the problems with Linux dump and ext2fs corruption. Is it true that XFS with the xfsdump tools ought to be free from these sorts of buffer cache issues? Or is it safest on Linux to stick with tools like gnutar which access the FS at a higher level? Here's the scoop from Nathan Scott, our userspace guru: Yes, I also read this thread - it was quite interesting. Though it wasn't explicitly spelt out, I think the problem that was being refered to with dump (not xfsdump) is that it looks at the on-disk metadata structures (via libext2fs) to figure out what needs dumping, etc. tar, et al. use the standard system call interfaces for traversing metadata and obtaining file data, so aren't exposed to the problems associated with dump-on-a-mounted-filesystem. xfsdump uses an xfs-specific ioctl (bulkstat) to walk the filesystem inodes, and doesn't refer directly to the on-disk data structures either, so it should be as safe as tar. HTH, -Eric -- Eric Sandeen XFS for Linux http://oss.sgi.com/projects/xfs [EMAIL PROTECTED] SGI, Inc.
Re: custom barcode labels
Hi Ron, I have a Brother P-touch 540 Extra that I use for making tape labels which also does several types of barcodes. However, someone else will have to verify whether the P-touch barcodes work with a changer or not because my changers don't read barcodes. I find using the P-touch a whole lot easier than printing out a whole sheet of labels from a printer, trying to find the right size labels or cut them to size, and trying to get the layout right. The only thing I don't like about it is that 9mm is the narrowest label size which is a little too wide for a 4mm tape. Perhaps Brother has some other model that supports a narrower width. You can check it out on www.brother.com Ron Snyder wrote: For those of you with tape changers and barcode readers: are any of you using any software to generate your own custom barcodes? It seems like it might be a bit nicer (for locating/moving tapes around) if the barcode label could match what the tape label has. I've only tried two pieces of software so far (BarCode Maker 3.0) and something called barcode 97 for windows 3.1. They both create barcodes OK, but for some reason my Qualstar tape library won't read them. (Maybe I'm doing something wrong, though-- I'm printing them out on regular paper, and then taping them to a test tape to see if the bar code reader can scan it. -- Jonathan F. Dill ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill
Linux SCSI utilities
Hi folks, I just found out about these sg_utils which may be helpful for folks running amanda on Linux systems, especially for debugging tapedrive and changer problems... http://gear.torque.net/sg/#Utilities: sg_utils and sg3_utils -- "Jonathan F. Dill" ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill Title: The Linux SCSI Generic (sg) Driver The Linux SCSI Generic (sg) Driver Introduction Background Features SG device driver downloads Utilities: sg_utils and sg3_utils Sg related pages External references SG in Redhat 6.2, 7.0 and Mandrake 7.1 Introduction The Linux sg driver is a upper level SCSI subsystem device driver that is used primarily to handle devices _not_ covered by the other upper level drivers: sd (disks), st (tapes) and sr (CDROMs). The sg driver is used by scanners, cd writers and applications that read cd audio digitally. Sg can also be used for less usual tasks performed on disks, tapes and cdroms. Sg is a character device which, in some contexts, gives it some advantages over sd and sr which are block devices. The interface of sg is at the level of SCSI command requests and their associated responses. Background The original driver was written by Lawrence Foard in 1992 and remained unchanged for several years. In August 1998 Heiko Eissfeldt and Joerg Schilling started working on enhancements to this driver. Soon after, the author became involved and these efforts culminated in a new sg driver being placed in Linux kernel 2.2.6 which was released on 16th April 1999. It contains the first major upgrade to the SCSI generic packet device driver ("sg") since 1992. This new driver has a super-set of the original interface and the semantics of the implementation are very similar. Hence it offers a high degree of backward compatibility with the original driver. The major reason for introducing a new sg driver into the 2.2 series of kernels was the problem that the original device driver was having trouble finding memory. This driver improves the situation by using scatter gather, memory above the 16 Mbytes level and memory from the scsi dma pool as appropriate. Other drivers were affected by these memory problems (especially those associated with ISA hardware). In kernel 2.2.10 H.J. Lu introduced a new kernel memory allocator that alleviated many of these memory problems. On 4th January 2001 the Linux 2.4.0 kernel was introduced and it contained a "version 3" sg driver that is described below. Features The following enhancements have been added: scatter gather, command queuing, per file descriptor sequencing (was per device) and asynchronous notification. Scatter gather allows large buffers (previously limited to 128 KB on i386) to be used. Scatter gather is also a lot more "kernel friendly". The original driver used a single large buffer which made it impossible to run 2 or more sg-based applications at the same time. With the new driver a buffer is reserved for each file descriptor guaranteeing that at least that buffer size will be available for each request on the file descriptor. A user may request a larger buffer size on any particular request but runs the (usually remote) risk of an out of memory (ENOMEM) error. A "version 3" sg driver is now available. It adds a new interface that allows more control over SCSI commands and returns more information about their performance. This driver is present in Linux kernel 2.4.0 . A separate version with reduced capabilities is available for the 2.2 series kernels. Features include: larger sense buffer, residual DMA count, 16 byte commands, direct IO support, command duration timing and a "proc_fs" interface. Naturally it is backward compatible with applications based on the sg interface in the lk 2.2 series and earlier. SG device driver downloads The following table summarizes the different versions of the sg device driver that are available. If you wish to use one of these tarballs then untar it in /usr/src/linux (or wherever the top of your kernel tree is). As a precaution you may wish to copy the files include/scsi/sg.h and drivers/scsi/sg.c to other names. This will facilitate reversing the patch if required. For information about the differences between versions see the history section at the top of the include/scsi/sg.h file. Linux version Sg driver version+tarball Patch Notes 2.2.6 sg22orig.tgz found in RH6.0, original documentation 2.2.6 + 2.2.7 sg2131.tgz pat_sg2134_225.gz 2.2.8 + 2.2.9 sg2132.tgz 2.2.10 - 2.2.13 sg2134.tgz patch 2.2.14 + 2.2.15 sg2136.tgz patch on sg version 2.1.34 (for documentation see bullet list) 2.2.16 sg2138.tgz patch on sg in lk 2.2.14+15 fixes scatter gather bugs on aha1542 2.2.17 - 2.2.19 sg2139.tgz patch on sg in lk 2.2.16 sg2138 broke sym53c416, now fixed optional 2.2 series sg version 3.0.17 sg version 3 backported to lk 2.2 optional 2.4 series sg version
Re: Amanda with two tape devices
Alexandre Oliva wrote: Can amanda use who tape devices to perform a single backup? You can't use them concurrently (yet), but you can set up chg-multi to switch between tape drives automatically. That's what we do here. Actually, there is a way that you can use them concurrently--You could split up your dumps into 2 dump configs, one dump config for each drive, which is what I have done. However, all of the disks for one system must be in the same dump config. You cannot have some disks from a system in one config, and other disks from that same system in the other config, or at least it would be extremely difficult to do that. The reason for this is that the client will already be busy with the first dump config, so when the second dump config runs, it will not be able to connect. If you run 2 dump configs concurrently, it is also a good idea to have separate holding disks for each config, or to split the amount of disk space that each config is allowed to use. For example, if you have an 18 GB holding disk, only let each config use 9 GB, otherwise disk space could run out unexpectedly and it might not be handled gracefully by amanda.. I think technically it should not be a problem, but in my experience I know that it has caused my backups to fail when both dump configs were writing to the same disk and used up all of the space. -- "Jonathan F. Dill" ([EMAIL PROTECTED])
Re: Advice: NFS vs SMB - looking for the voice of experience
Alexandre Oliva wrote: I'd much rather use NFS than SMB. It's generally far more efficient. However, God only knows how much crap an NFS server running on MS-Windows would have to work against, so it might be that it actually takes longer to run. I recommend running some I/O benchmarks eg. bonnie with a 100 MB or 256 MB file over NFS then over SMB. My experience has been that Sun PCNFS is incredibly slow, but some other NFS implementation on NT might be faster. -- "Jonathan F. Dill" ([EMAIL PROTECTED])
Re: Advice: NFS vs SMB - looking for the voice of experience
This point is very important. You will have to do the equivalent of exporting to the server with "root" enabled. In Unix this usually is an option like "root=X" or on Linux "no_root_squash" otherwise you may not have sufficient priviledges to read the files. It may look like the backups worked, but when you restore the files, you may find that the files are the right size but only contain null characters (aka. ^@ or ASCII 0). It all depends how the MS NFS implementation handles UID mapping and what happens when you have insufficient priviledges to access some file. If you choose to use this NFS arrangement, you should make sure to export the disk read-only, otherwise someone could use NFS to trash your NT server(s). You should also try restoring a backup to a different location, eg. the holding disk, and make sure the file contents are OK and not bogus ^@ files. "John R. Jackson" wrote: ... I lean towards NFS, is there any reason I should not? I know very little about this, but the one thing that popped to mind is whether an MS NFS server would give a tar running as root on a client (to NFS) enough access to get to everything. The normal action is to convert all root requests to "nobody", which will not work well for backups. -- "Jonathan F. Dill" ([EMAIL PROTECTED])
Re: Error: gtar: uid_t value 4294967294 too large (max=16777215)
Hi Terry, I have seen this problem with Unix computers running either Samba, Appletalk sharing, or PCNFS. Something, possibly a misconfigured Samba, is probably using that UID as the "nobody" UID. If you're not using Samba for anything and it's just installed and "turned on" I think it would be a good idea to turn it off as it could be a security liability, plus that may take care of your UID problem. You also might try upgrading to tar-1.13.19 if you haven't already--that seems to clear up some problems where an exit code should be returned to indicate a "warning" rather than a total "failure." Terry Koyama wrote: I ran amanda last night and received the below report. The client with the error (hendrix) is an AIX machine running amanda v2.4.2p1. I tried checking out the UID's in /usr but didn't find anything out of the ordinary. I've also included sendback.debug and sendsize.debug to see the actual command that was executed. The syntax of runtar looks kinda funky (to me anyway) but seems to run fine on all the other clients. Plus, I don't have a clear understanding of how runtar and gnutar are related. -- "Jonathan F. Dill" ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill
Re: Error: gtar: uid_t value 4294967294 too large (max=16777215)
"John R. Jackson" wrote: I checked the /etc/passwd and /etc/group files which had user/group nobody with the id of 4294967294. After I changed that ... Ummm, you changed the gid and uid of "nobody"? That was probably kind of rash. There are things floating around that know about that and expect a specific value. Like NFS :-). Yes, for example we have a piece of lab equipment from England which requires an Acorn computer. If you don't have other Acorn devices on the network, usually the only way to print and share files is via PCNFS, and to get that to work correctly I had to set UID nobody to some such value as that, and changing the value would "break" printing and file sharing from the Acorn. -- "Jonathan F. Dill" ([EMAIL PROTECTED])
huge incrementals
Hi all, What values are people using for bumpsize, bumdays, bumpmult other than the example values? I have a lot of 10 GB disks to back up and it doesn't seem efficient to me to do eg. an 11 GB "level 5" backup of a disk that has 13 GB of data on it--In that situation, I think it would probably make more sense to just do a level 0 backup. On the other hand, I don't want to create a situation where disks 4 GB get a full dump every time. It would be nice if bumpsize could be expressed as a % of the disk size rather than an absolute amount of space. Any hints anybody can give me in tuning the bump parameters would be appreciated. Thanks, -- "Jonathan F. Dill" ([EMAIL PROTECTED])
Re: using a DDS-4 Autoloader under linux
I'm using a Sony TSL-11000 DDS4 autoloader under Linux without problems. It works as an Ultra2 LVD device without problems. I think I'm getting better throughput than 2.8 MB/s, but it's still well under Fast/Narrow bandwidth due to the limitations of the drive. I use mtx and random access works as well as sequential. I had to modify the chg-mtx script slightly--I can send it to you if you get the TSL-11000. The one problem that I have had is an occasional tape getting stuck during the change process, and that's using Sony brand DDS4 tapes and not some cheap imitation. 3 times so far I had to take the cover off the drive, mark the stuck tape as "bad," and give it a little push back into the magazine. Then you have to shut down the system and power cycle the drive to get the error to reset and magazine to eject. I don't know if any of the other autoloaders ever have this problem. Martin Apel wrote: On Thu, 8 Mar 2001, Werner Behnke wrote: We are using a Seagate Scorpion DDS-4 Autoloader without problems. The only issue was, that for an unknown reason it doesn't like to be connected to the SCSI controller as a 'wide' device. If you set the DIP switches, such that it will register itself as a 'narrow' SCSI device everything is fine. Performance is about 2.8 MB/s so the narrow cable should not slow down the device noticeably. Does the autoloader support random access or only sequential mode? It supports random access. Did you set up chg-manual in amanda to change tapes manually? If yes: how do you control the media changer? With mtx (http://mtx.sourceforge.net/) or Autoloader Tape Library (http://www.ee.ryerson.ca/~sblack/autoloader/) or something else (mt, SCSI UNLOAD command, Amanda's chg-mtx...)? I took chg-mtx and adapted it at one place to use mtx 1.2. Using Linux for the tape server you have to configure the kernel with multiple LUNs per SCSI device. Greetings, Martin -- "Jonathan F. Dill" ([EMAIL PROTECTED])
Re: Don't open attachment!!!
"Anthony A. D. Talltree" wrote: IMHO, anyone who insists on using the software that's vulnerable to such attacks deserves to lose. OTOH the amanda-users list doesn't deserve to lose if someone is dumb enough to open the attachment, gets infected, and sends all kinds of crap back to the list. After all, the message probably got here in the first place from someone on amanda-users who got infected. Anyway, here is the official dirt from SARC: http://www.sarc.com/avcenter/venc/data/w95.hybris.gen.html -- "Jonathan F. Dill" ([EMAIL PROTECTED])
[Fwd: enp3.unam.mx spam relay]
Hello all, My experience trying to help some scientists at UNAM is that unam.mx is a disaster area, therefore this e-mail is a shot in the dark at best. FYI I am not an admin of amanda-users or the server that hosts it, so don't complain to me about spam or other problems. Original Message Subject: enp3.unam.mx spam relay Date: Mon, 19 Feb 2001 14:35:34 -0500 From: Jonathan Dill [EMAIL PROTECTED] Organization: CARB To: [EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED] Dear administrators, enp3.unam.mx has incorrect e-mail configuration which allows spammers everywhere in the world to use it to send spam. Please refer the admin of the system to http://maps.vix.com/tsi for instructions on the correct way to configure a SMTP server. This particular spammer sends several spam messages to the amanda-users-list which is very rude and annoying, so I hope somebody can do something about it. Original Message Return-Path: [EMAIL PROTECTED] Received: from ashd1-1.relay.mail.uu.net ([199.171.54.245]) byumbi3.umbi.umd.edu (Netscape Messaging Server 4.15) with ESMTPid G90R4300.HKS for [EMAIL PROTECTED]; Mon, 19 Feb 2001 14:21:39 -0500 Received: from surly.omniscient.com by mr0.ash.ops.us.uu.net with ESMTP (peer crosschecked as: surly.omniscient.com [208.213.83.10])id QQkdbx04110;Mon, 19 Feb 2001 19:18:09 GMT Received: (from root@localhost)by surly.omniscient.com (8.11.1/8.11.1) id f1JJE9a169507for amanda.org.amanda-users-list; Mon, 19 Feb 2001 19:14:09 GMT Received: from enp3 (enp3.unam.mx [132.248.93.100])by surly.omniscient.com (8.11.1/8.11.1) with SMTP id f1JJCld169568for [EMAIL PROTECTED]; Mon, 19 Feb 2001 14:12:47 -0500 (EST) Received: from 132.248.93.100 by enp3 (SMI-8.6/SMI-SVR4)id UAA03664; Thu, 15 Feb 2001 20:20:14 -0600 From: [EMAIL PROTECTED] Message-Id: 200102160220.UAA03664@enp3 To: [EMAIL PROTECTED] Date: Thu, 15 Feb 01 01:26:44 EST Subject: toner supplies Sender: [EMAIL PROTECTED] Precedence: bulk PLEASE FORWARD TO THE PERSON RESPONSIBLE FOR PURCHASING YOUR LASER PRINTER SUPPLIES VORTEX SUPPLIES -SPECIALS OF THE DAY ON LASER TONER SUPPLIES AT DISCOUNT PRICES-- LASER PRINTER TONER CARTRIDGES COPIER AND FAX CARTRIDGES WE ARE --THE-- PLACE TO BUY YOUR TONER CARTRIDGES BECAUSE YOU SAVE UP TO 30% FROM OFFICE DEPOT'S, QUILL'S OR OFFICE MAX'S EVERY DAY LOW PRICES ORDER BY PHONE:1-888-288-9043 ORDER BY FAX: 1-888-977-1577 CUSTOMER SERVICE: 1-888-248-2015 E-MAIL REMOVAL LINE: 1-888-248-4930 UNIVERSITY AND/OR SCHOOL PURCHASE ORDERS WELCOME. (NO CREDIT APPROVAL REQUIRED) ALL OTHER PURCHASE ORDER REQUESTS REQUIRE CREDIT APPROVAL. PAY BY CHECK (C.O.D), CREDIT CARD OR PURCHASE ORDER (NET 30 DAYS). IF YOUR ORDER IS BY CREDIT CARD PLEASE LEAVE YOUR CREDIT CARD # PLUS EXPIRATION DATE. IF YOUR ORDER IS BY PURCHASE ORDER LEAVE YOUR SHIPPING/BILLING ADDRESSES AND YOUR P.O. NUMBER NO SHIPPING CHARGES FOR ORDERS $49 OR OVER ADD $4.75 FOR ORDERS UNDER $49. C.O.D. ORDERS ADD $4.5 TO SHIPPING CHARGES. FOR THOSE OF YOU WHO REQUIRE MORE INFORMATION ABOUT OUR COMPANY INCUDING FEDERAL TAX ID NUMBER, CLOSEST SHIPPING OR CORPORATE ADDRESS IN THE CONTINENTAL U.S. OR FOR CATALOG REQUESTS PLEASE CALL OUR CUSTOMER SERVICE LINE 1-888-248-2015 OUR NEW (LASER PRINTER TONER CARTRIDGE) PRICiES ARE AS FOLLOWS: (PLEASE ORDER BY PAGE NUMBER AND/OR ITEM NUMBER) HEWLETT PACKARD: (ON PAGE 2) ITEM #1 LASERJET SERIES 4L,4P (74A)$44 ITEM #2 LASERJET SERIES 1100 (92A)-$44 ITEM #3 LASERJET SERIES 2 (95A)---$39 ITEM #4 LASERJET SERIES 2P (75A)-$54 ITEM #5 LASERJET SERIES 5P,6P,5MP, 6MP (3903A)--$44 ITEM #6 LASERJET SERIES 5SI, 5000 (29A)--$95 ITEM #7 LASERJET SERIES 2100 (96A)-$74 ITEM #8 LASERJET SERIES 8100 (82X)---$145 ITEM #9 LASERJET SERIES 5L/6L (3906A0--$35 ITEM #10 LASERJET SERIES 4V-$95 ITEM #11 LASERJET SERIES 4000 (27X)-$72 ITEM #12 LASERJET SERIES 3SI/4SI (91A)$54 ITEM #13 LASERJET SERIES 4, 4M, 5,5M---$49 HEWLETT PACKARD FAX (ON PAGE 2) ITEM #14 LASERFAX 500, 700 (FX1)--$49 ITEM #15 LASERFAX 5000,7000 (FX2)--$54 ITEM #16 LASERFAX (FX3)$59 ITEM #17 LASERFAX (FX4)$54 LEXMARK/IBM (ON PAGE 3) OPTRA 4019, 4029 HIGH YIELD---$89 OPTRA R, 4039, 4049 HIGH YIELD-$105 OPTRA E$59 OPTRA N--$115 OPTRA S--$165 - EPSON (ON PAGE 4) ACTION LASER 7000,7500,8000,9000---$105 ACTION LASER 1000,1500-$105 CANON PRINTERS (ON PAGE 5) PLEASE CALL FOR MODELS
Re: spam messages in the amanda-user mailing list
Ryan Williams wrote: There are now daily spam messages about toner supplies going to the amanda mailing list. This is a big annoyance. Please do something to prevent such a thing from happening again. If needed I can provide headers and the emails that I recieved. I'm mad about it too, but what makes you think that John R. Jackson can do anything about it, or anybody else subscribed to the list for that matter? Majordomo doesn't have much capabilities for spam filtering as far as I know. Personally, I'd like to see the list run on mailman rather than majordomo. The list appears to be hosted on surly.omniscient.com. I'm going to inquire about helping out with the admin of the server and some other options. Until then, I'd suggest getting a real e-mail client like Netscape Messenger and use a message filter on "Sender-Contains" and "toner" then "Move To-Spam" or something like that. -- "Jonathan F. Dill" ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill
Re: amlabel + rack of tapes question
Hi Joe, Writing a script is a lot easier than you might think--I just did it to label about 90 tapes using a DDS4 autoloader with 8 tape magazine, it was just a 14 line tcsh script with almost no debugging involved. The script is very site-specific since it was a one-off type of project, but I attached it anyway so you can get some ideas. My script labels one magazine full of tapes, then you have to reload the magazine and run the script again. It wouldn't be too hard to change the script to use the correct amanda API like using amtape and getting info from amanda.conf file, but I haven't got the time or motivation to do it right now. If you're in the middle of your dump cycle and want to add some tapes onto the end of the dump cycle, the tricky part is fixing your tapelist file--If you don't fix your tapelist, on the next run amanda will want to use the tapes that you just labeled instead of continuing the current cycle. The easy way to fix it is to take the date stamp from the last tape in the cycle and change the "0" for all of the new tapes to that date--and of course don't forget to adjust tapecycle and other parameters in amanda.conf--and then amanda should finish the current cycle before starting to use the new tapes. Joseph Del Corso wrote: Is it possible to amlabel an entire rack of tapes without doing it manually for each tape? Specifically I have roughly 35 tapes that I'd like to label in some kind of automated fashion all at once. Besides writing my own script (which would take time and more than likely a HECK of a lot of debugging) is there an easier way to do this? Is there any benefit to doing this? -- "Jonathan F. Dill" ([EMAIL PROTECTED]) CARB Systems and Network Administrator Home Page: http://www.umbi.umd.edu/~dill #!/bin/tcsh -f set i=2 while ($i 9) set j=`awk '{split($2,a,"."); print a[2]}' dds4/tapelist | sort -n | tail -1` @ j++ amlabel dds4 dds4.$j mtx -f /dev/sg3 unload mtx -f /dev/sg3 load $i @ i++ end set j=`awk '{split($2,a,"."); print a[2]}' dds4/tapelist | sort -n | tail -1` @ j++ amlabel dds4 dds4.$j mtx -f /dev/sg3 eject
Re: Strange dump summary...
Hi Suman, Did you check on the client to see what's in the debug files of the /tmp/amanda directory? Check the permissions on /usr/local/libexec/runtar--it must be owned by root and SUID like this: -rwsr-x---1 root amanda 78334 Nov 13 15:32 /usr/local/libexec/runtar I had a problem recently where some genius decided to something like: chown -R foobar.silly /usr/local and backups of his machines stopped working because of it. If possible, you may just want to run "make install" again to make sure that the permissions of everything are set correctly. I imagine that it's also possible that you could have some sort of security patched kernel which blocks SUID execution which would stop runtar from working--I vaguely remember hearing of some sort of kernel patch like that, but it wouldn't be in some stock kernel eg. from Red Hat or another major distribution. Suman Malla wrote: Hi there, Could someone pls enlighten me? I am trying to backup a filesystem on redhat server (mogadon) but in vain. amcheck doesn't complain at all though. Amanda is backing up 8 other servers without any problem for the last 4-5 months but with this server somehow it fails. Messages from log files - Amanda report: = FAILURE AND STRANGE DUMP SUMMARY: mogadon/etc lev 0 FAILED [disk /etc offline on mogadon?] [snip]... mogadon /etc 0 FAILED [snip]... amdump.1: setting up estimates for mogadon:/etc mogadon:/etc overdue 11355 days for level 0 setup_estimate: mogadon:/etc: command 0, options: last_level -1 next_level0 -11355 level_days 0 getting estimates 0 (0) -1 (-1) -1 (-1) got result for host mogadon disk /etc: 0 - -1K, -1 - -1K, -1 - -1K FAILED QUEUE: 0: mogadon/etc Any hint, point would be greatly appreciated. TIA. - smallA ___ Send a cool gift with your E-Card http://www.bluemountain.com/giftcenter/ -- "Jonathan F. Dill" ([EMAIL PROTECTED])