Re: A few interesting problems (tapeless, interactive restores, etc)
Hi John, I am looking at amanda-2.4.3b3, but aside from docs/VTAPE-API I don't see any mention of how the virtual tapes can be set up with AMANDA. And the VTAPE-API looks to me, as a non-coder, like possibly more than I need to know (though if i need to hack some C code I can). This sounds like it would be much easier for me, as having a virtual tape name to ask of ADSM is much easier than trying to figure out which tapes I need on my own, and restoring a dump disk for the purpose. Can you point me to any documentation, crude or otherwise, describing this process? If anyone else has gotten this to work, what does it involve? If I need to change tapeio.c, use special amanda.conf directives, etc etc etc (nothing is mentioned in sample files). I appreciate any assistance. Greg On Wed, 2002-04-03 at 22:39, John R. Jackson wrote: First off, for reasons beyond the scope of this email, my AMANDA server has no tape device. I am sending all backups to the holding disk. We also have a large IBM tape library running ADSM that I would like to incorporate into my AMANDA backup strategy. ... Sounds like a perfect setup for the tapeio code in 2.4.3 (now in beta test, but the tapeio stuff has been stable for a long time). You would set up a disk area to emulate a tape, then use the chg-multi tape changer to move the data back and forth with your ADSM system. The current (2.4.3 beta) chg-multi has a posteject hook to a script you provide that could move the date into ADSM. It would be easy to add a preload step to do the other direction (don't know why I didn't think of doing that when I did posteject). A) I need a way of asking AMANDA, If I want foo file from 3/20/2000 on machine bar, which dump 'directories' will I need? ... The tapeio code would take care of this. Amanda would tell you you needed tapes A, B and C, which would map directly to your ADSM areas. If you don't go with tapeio, then I think amadmin config find ... is what you'll want. You may also need to be a bit sneaky about the data motion to ADSM and leave the names and holding disk directories behind, but truncate the files to zero length, or, worst case, truncate them to just their 32 KByte header. I'm pretty sure that would fool amrecover sufficiently into using them (obviously, you'd have to reload the rest of the data before turning the restore completely loose). Greg Mohney John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED] -- Greg Mohney Technical Specialist, UNIX Administration APAC Customer Services (319) 896-5027
Re: A few interesting problems (tapeless, interactive restores, etc)
On 4 Apr 2002 at 11:04am, Greg Mohney wrote Can you point me to any documentation, crude or otherwise, describing this process? If anyone else has gotten this to work, what does it involve? If I need to change tapeio.c, use special amanda.conf directives, etc etc etc (nothing is mentioned in sample files). I appreciate any assistance. Documentation on how to use tapeio is in the amanda(8) man page shipped with 2.4.3bN. Look for stuff about specifying file: as your tapedev. -- Joshua Baker-LePain Department of Biomedical Engineering Duke University
A few interesting problems (tapeless, interactive restores, etc)
Hello, I am implementing AMANDA at my current workplace (used it once before and loved it) however I have a few strange circumstances and questions regarding my options. If anyone can help me with the below it would be appreciated. First off, for reasons beyond the scope of this email, my AMANDA server has no tape device. I am sending all backups to the holding disk. We also have a large IBM tape library running ADSM that I would like to incorporate into my AMANDA backup strategy. In my head, it will go something like this: 1. AMANDA schedules and performs backups to holding disk. 2. Every night, after AMANDA has completed her backups, the dated directory is moved over to our ADSM server. 3. The holding disk is wiped clean (space saving measure, ADSM will have the files anyway). 3. The ADSM server spools the directory to tape. 4. I write a wonderfully sexy amrestore-wrapper script that will take three arguments: * machine name * files to be restored * date of file you want to extract 5. Wrapper script figures out which dump directories it needs to get out of ADSM, gets them, and then runs the appropriate amrestore commands to seemlessly restore said files. I have a couple of problems (i.e. things I'm not sure about or are too stupid to understand) however and they are as follows: A) I need a way of asking AMANDA, If I want foo file from 3/20/2000 on machine bar, which dump 'directories' will I need? I have read and re-read the man pages for amdump, amadmin, etc., but the answer is not immediately obvious to me. I understand AMANDA usually works with tapes, but since even amrecover will look into holding disk dated directories, surely there's a way to get this info from AMANDA (without actually performing the restore, all I want is the dated directory names so I can pull them out of ADSM and slap them back into the AMANDA server's holding disk, so that amrestore/amrecover can get to them). B) After writing the above, I'm not sure I remember what my second problem was, or if it was relevent if (A) is solved. Sorry. I think it had something to do with piping 'amrestore' output to tar not being interactive as it could be with dump/restore. In other words, I want our tape monkies to be able to interactively restore files, but if I'm using GNU tar all I can do is request specific files. Eh, but that's ok I think...sorry... Anyways, I know this is long and rambling. I appreciate any help on the above, or even different strategies if the one presented is on crack. Thanks. -- Greg Mohney Technical Specialist, UNIX Administration APAC Customer Services (319) 896-5027
Re: A few interesting problems (tapeless, interactive restores, etc)
First off, for reasons beyond the scope of this email, my AMANDA server has no tape device. I am sending all backups to the holding disk. We also have a large IBM tape library running ADSM that I would like to incorporate into my AMANDA backup strategy. ... Sounds like a perfect setup for the tapeio code in 2.4.3 (now in beta test, but the tapeio stuff has been stable for a long time). You would set up a disk area to emulate a tape, then use the chg-multi tape changer to move the data back and forth with your ADSM system. The current (2.4.3 beta) chg-multi has a posteject hook to a script you provide that could move the date into ADSM. It would be easy to add a preload step to do the other direction (don't know why I didn't think of doing that when I did posteject). A) I need a way of asking AMANDA, If I want foo file from 3/20/2000 on machine bar, which dump 'directories' will I need? ... The tapeio code would take care of this. Amanda would tell you you needed tapes A, B and C, which would map directly to your ADSM areas. If you don't go with tapeio, then I think amadmin config find ... is what you'll want. You may also need to be a bit sneaky about the data motion to ADSM and leave the names and holding disk directories behind, but truncate the files to zero length, or, worst case, truncate them to just their 32 KByte header. I'm pretty sure that would fool amrecover sufficiently into using them (obviously, you'd have to reload the rest of the data before turning the restore completely loose). Greg Mohney John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
