Re: amrecover: Why does it use disklist and log files
Hello Jean- Jean-Louis Martineau wrote: Hi Toralf, On Wed, Oct 22, 2003 at 10:09:56AM +0200, Toralf Lund wrote: As far as I can't tell, amrecover won't work unless 1. Log file from the backup you are trying to recover is still present The log are needed to find which dump are on which tape and where on that tape. We have the need to keep both weekly full and daily incrementals on disk (tapeless) so that we won't have to go to slow media. We then copy full backups (data and inex) to magnetic tapes for a rotation of 8 tapes using tar. The disk won't hold up to 8 full backups, so it is set up with only one "tape". After full backup, data and index are copied immediately to tape. This way, we can go back one week, up to any day, using data on disk; and go back every week before that, using data on tapes. We do not care much about previous week's incrementals. Recover using tapes is done, first with untarring it from tape onto disk and replacing necessary index (with data on disk renamed first for preservation, of course), then running amrecover. My questions is: Must I have log files tarred to tape along with data and index in order to run amrecover? I have never needed to go back more than one week from tape. Thanks in advance.
Re: amanda skipped two runs.
Thanks Joshua and Darin for the pointers. I'm wondering if anyone has done a kernel upgrade to 2.4.20-20.9 (RH9.0) as mentioned that it fixes signal delivery race condition, as I don't know whether to follow the Amanda patch, or the kernel patch. Joshua Baker-LePain wrote: On Mon, 22 Sep 2003 at 11:01am, Bao Ho wrote RedHat Linux 9.0, kernel 2.4.20-18.9 Amanda 2.4.4p1-20030716 Server only backs up itself. No suprious email, ie cron output ? I reported in my subsequent email. I see two unusual messages in syslog: application bug: dumper (PID) has SIGCHLD set to SIG_IGN but callls wait(). (see the NOTES section of 'man 2 wait'). Workaround activated. Those messages are due to the NPTL patches in the RH9 kernel. Look through the list archvies for a patch from someone at RH fixing this problem.
Re: amanda skipped two runs.
First, sorry for the wrong indentation. Here's the info I just found. I thought it's just a quick question, turns out there may be more Brian Cuttler wrote: At my shop the amanda tapes are labeled MACHINxx and we use the tapes in (assending) numeric order. If a run gets skipped and it wants the next tape we actually maintain the ordering. How are your tapes labeled ? My virtual tapes (back up to disk) are labeled DailySet1-6, and correspond to Wed-Mon. They are used in that fashion also. Cron (well, the one I'm using) keeps a log file. You are right, syslog (messages? what OS are you running?) file may also be of help. RedHat Linux 9.0, kernel 2.4.20-18.9 Amanda 2.4.4p1-20030716 Server only backs up itself. No suprious email, ie cron output ? I reported in my subsequent email. I see two unusual messages in syslog: application bug: dumper (PID) has SIGCHLD set to SIG_IGN but callls wait(). (see the NOTES section of 'man 2 wait'). Workaround activated. Then, the server seems to be rebooted by someone at our place. It must have been a crash and then a hang that caused someone to do that. ======== Bao Ho wrote: Hi List, I just found out the Amanda skipped two days' worth of backup on last Sat and Sun. It now insists on using the tape for Sat as the next tape for today's run. Is there any way I can force it to use the correct tape for today w/o messing up the cycle? Is there anything I can do to figure out why the backup didn't run? I'm looking at syslog to see if anything happened before Sat. Thanks
amanda skipped two runs.
Hi List, I just found out the Amanda skipped two days' worth of backup on last Sat and Sun. It now insists on using the tape for Sat as the next tape for today's run. Is there any way I can force it to use the correct tape for today w/o messing up the cycle? Is there anything I can do to figure out why the backup didn't run? I'm looking at syslog to see if anything happened before Sat. Thanks
Re: Rewinding tapeless "tapes": why? error?
>> >> That is what I used Paul >> ammt -f rewind. > Okay, but if you specified /dev/null as your tapedev in amanda.conf, > you are *not* using disk-pseudo-tapes; you are simply leaving files > in your staging area. > > If you were using disk-pseudo-tapes, you would have a tapedev of > "file:/some/directory/name" and "ammt -f file:/some/directory/name rewind" > would work fine. Marc, I don't explicitly specify /dev/null as my tapedev. It defaults to that value. I followed some instructions I found from the archive that for tapeless backup, you do not need to specify tapedev. I do have these followings: tpchanger "chg-multi" changerfile "/.../changer.conf" the changer.conf has: multieject 0 gravity 0 <...snip...> firstslot 1 lastslot 6 slot 1 file:/path/to/tape/file1 slot 2 file:/path/to/tape/file2 <...snip...> I only did "ammt -f /path/to/tape/file1/1.server._home.1 rewind" I tried using "ammt -f file:/path/to/tape/file1/1.server._home.1 rewind" as you suggested, and this is the error I got: file:/path/to/tape/file1/1.server._home.1 rewind failed: Not a directory Any other ways I may use??
Re: Backup plan needs advice.
This is getting longer and longer. So, I am going to reply to both Jon's and
Gene's emails in this one email.
-Original Message-
From: Jon LaBadie <[EMAIL PROTECTED]>
To: Amanda Users (E-mail) <[EMAIL PROTECTED]>
Date: Tuesday, June 24, 2003 2:57 PM
Subject: Re: Backup plan needs advice.
>On Tue, Jun 24, 2003 at 02:25:20PM -0700, bao wrote:
>>
>> Jon LaBadie wrote:
>>
>> >strategy "string"
>> >
>> > ...
>> > incronly
>> > Only do incremental dumps. `amadmin force' should
>> > be used to tell Amanda that a full dump has been
>> > performed off-line, so that it resets to level 1.
>> > It is similar to skip-full, but with incronly full
>> > dumps may be scheduled manually. Unfortunately,
>> > it appears that Amanda will perform full backups
>> > with this configuration, which is probably a bug.
>> >
>> >
>> >Also, another dumptype option entry was this paragraph:
>> >
>> >skip-full "boolean"
>> > Default: no. If true and planner has scheduled a full
>> > backup, these disks will be skipped, and full backups
>> > should be run off-line on these days. It was reported
>> > that Amanda only schedules level 1 incrementals in this
>> > configuration; this is probably a bug.
>> >
>> >
>> >
>> As pointed out. The two paragraphs above have "a bug". I prefer not to
>> mess with those options. :)
>
>Well, they say "probably". Maybe someone was saying this program/feature
>works this way. Probably not only way it could have been designed
>and maybe not the best choice for the design -- thus a "bug".
>Example, according to the man page for standard unix sort command
>has been the line "silently truncates lines longer than bytes".
>It could be called a "bug". But in design and implementation,
>not a defect in coding.
>
>
>Plus, the comments may be a holdover from early releases never updated.
>There are comments a few years back suggesting the "bug" was fixed.
>
>
>BTW you could also consider having your "full config" dump to real tape
>or to a separate set of "file tapes". Then you could run your daily
>config as a normal amanda config doing a mixture of full and incrementals
>each day. If you run the full dump to real tape and the clerical person
>forgets the tape, it can collect on the holding disk for later amflush.
>Otherwise you will have to develop your own procedures for moving the
>dumps and the indexes and the ... to tape and later, if needed, you own
>procedures for recovering them and the indexes and ... from the tape
>for am{recover|restore}.
>
As I have two configs and two sets of "tapes" now, but they behave as just
one continuous set of tapes.
If I do as Jon said, dump the full to a real tape, and the inc to a set of
"tapes", then the full to tape and
the full of the daily config probably will not be consistent, as they will
have to be run at different times.
I think someone had given his/her instructions of how to copy the backup
image and the index some while back
(maybe Gene) to tape. I will have to dig more into my collection of mail
from this group to look for it.
I am still leaving it to run with my original two configs and waiting to see
what will happen in the next few days
as the new cycle begins. If it doesn't work out, I will look into what Jon
suggested, one config.
I'm still facing the problem of verifing if the next first incremental will
follow the next full, or the full from the previous
week. Please give me any suggestions you can think of to accomplish this.
Gene, diskspace is not a problem to us. Our data consists of only around 20
GB. We have _plenty_ plenty of hard disk
space (it's very cheap nowadays). Our tape is 20 GB native, expanding to 40
GB with compression.
It's been a very helpful and fun discussion. Many thanks to both Jon and
Gene.
I hope there will be follow-ups.
Cheers.
>--
>Jon H. LaBadie [EMAIL PROTECTED]
> JG Computing
> 4455 Province Line Road(609) 252-0159
> Princeton, NJ 08540-4322 (609) 683-7220 (fax)
>
Re: fixed day backup
Hi David, Jay, and Wayne, Thanks for replying. I had tried the options "strategy noinc" and "strategy nofull" as Jay has mentioned. In fact, we can get away with only one config with two dumptypes and have to call the correct one on the correct day with cron. We just override the dumpcycle options in the dumptypes. It works so far, but there was the Mail drive from another machine, which I use smb to mount to the backup server, that almost always has a full backup, no matter if it is called with the incremental dumptype or the full dumptype. Its contents changes constantly, but I don't think that's the reason why it should be fully backed up. I also tried two separate configs, in combination with the two options mentioned above. I also used amadmin to force a level 0 backup right before the full is performed. I will try with "record no" as soon as I get back to the office. Maybe because I was using all of your suggested methods at the same time that caused the problem :) -Original Message- From: David Olbersen <[EMAIL PROTECTED]> To: bao <[EMAIL PROTECTED]>; amanda <[EMAIL PROTECTED]> Date: Thursday, March 13, 2003 2:53 PM Subject: RE: fixed day backup >Bao, > >I don't know why the FAQ says that -- it's simply not true! > >We currently have a setup similar to this: incremental every day and full backups every other weekend. > >The trick is to make 2 amanda configurations (say, 'Daily' and 'Archive'). > >Once you've got that you can just call amanda with the correct configuration on the right days, i.e. > >amdump Daily >amdump Archive > >-- >David Olbersen >iGuard Engineer >11415 West Bernardo Court >San Diego, CA 92127 >1-858-676-2277 x2152 > > >> -Original Message- >> From: bao [mailto:[EMAIL PROTECTED] >> Sent: Thursday, March 13, 2003 12:25 PM >> To: amanda >> Subject: fixed day backup >> >> >> Hello, >> >> In the Amanda FAQ, there is one question "How can I configure >> Amanda to >> perform full backups on the weekend and incrementals on >> weekdays?". The >> answer is "You can't." >> >> It's unfortunate that the plan above is what I need. I need >> to run full >> backup once a week, and incrementals on the other days. All of the >> backups will stay on disk (tapeless backup), plus the full >> will be put >> on tape. Disk only keeps one week (full + incrementals), and >> tapes keep >> 8 weeks of full-only. >> >> If Amanda manages the full and incremental using its own way. >> How can I >> know which one has the full, and when the full is performed, >> to automate >> the process of transferring to tape?? >> >> Has anyone done this sucessfully and How can I do it?? >> >> Best regards, >> >> >
Re: EOF, check amidxtaped.debug file [..]
>On Mon, Mar 03, 2003 at 12:54:40PM -0800, bao wrote: >> I just got a similar problem to what Odd Rune Dahle described, with a >> slightly different message. >> >> In the amidxtaped..debug file, almost everything went fine until: >> >> amrestore: error reading file header: Is a directory >> amidxtaped: time 0.004: amrestore terminated normally with status: 2 >> amidxtaped: time 0.004: rewinding tape ... >> amidxtaped: time 30.102: tape_rewind tape: /home/amanda/testSet1: >> Inappropriate ioctl for device >> amidxtaped: time 30.102: pid 25439 finish blah blah blah >> >> 1. The "Inappropriate..." must be because I use tapeless back up on a >> second hard drive. > >What's your tapedev in amanda.conf? You should use it. >Maybe it's 'file:/home/amanda/testSet1'. > I don't have an entry for tapedev in amanda.conf. I use chg-multi for the changer, with changerfile "/usr/.../changer.conf" In changer.conf, there are these entries: slot 1 file:/home/amanda/testSet1 slot 2 file:/home/amanda/testSet2 ... >Jean-Louis >> >> 2. Is amrestore capable of restoring a whole directory tree??? >> And what goes wrong here? >> >> Thanks, >> >> >> Odd Rune Dahle wrote: >> >Joshua Baker-LePain: >> > >> >EOF, check amidxtaped.debug file on ns1. >> >> And did you do this? What's in that file? >> >>> >> >>>I'm sorry. I forgot to mention that this file doesn't exist. >> >> >> >>First off, what user you running that command as? If it's not root or >> >>amanda, then it's not going to pick up /tmp/amanda (where the debug files >> >>live), b/c /tmp/amanda has 700 permissions. >> > >> > >> >It was as root. But now that you mention /tmp/amanda, I checked there. >> >amrestore refer to amidxtaped.debug, but the file is named >> >amidxtaped..debug.. no wonder the locate command didn't >> >find it :-) >> > >> >Things made a bit more sense once I found the debug-file: >> > >> >amrestore: could not open /dev/tape: Permission denied >> >amidxtaped: time 0.003: amrestore terminated normally with status: 2 >> >amidxtaped: time 0.003: rewinding tape ... >> >amidxtaped: time 0.003: tape_rewind: tape open: /dev/tape: Permission >> >denied >> >amidxtaped: time 0.003: pid 10103 finish time Fri Feb 28 07:19:30 2003 >> > >> >My xinetd-setup specified the wrong group for amidxtaped (bin), when I >> >use 'disk' as owner of the tape-device. Unfortunately there are >> >nobody on-site until monday morning, but we'll run a new test then. >> > >> >Thanks for the help :-) >> > >> > >-- >Jean-Louis Martineau email: [EMAIL PROTECTED] >Departement IRO, Universite de Montreal >C.P. 6128, Succ. CENTRE-VILLETel: (514) 343-6111 ext. 3529 >Montreal, Canada, H3C 3J7Fax: (514) 343-5834 >
Re: holding disk and full dumps
-Original Message- From: Dietmar Goldbeck <[EMAIL PROTECTED]> To: bao <[EMAIL PROTECTED]> Cc: amanda <[EMAIL PROTECTED]> Date: Thursday, February 27, 2003 6:28 PM Subject: Re: holding disk and full dumps >On Thu, Feb 27, 2003 at 02:09:33PM -0800, bao wrote: >> >> 1. If I want to do a full dump of the entire selected folder every week, >> and incremental dumps every other days (1 full, 6 inc. per week), does >> it mean I have to rely on amanda in choosing what to do with load >> balancing?? >> > >Yes. This is one advantage Amanda has over commercial utilities. >She does all the scheduling and very well. You can force backups >if you want with amadmin force . >See manpage of amadmin. > >> 2. I read somewhere (which I can't find where and can't remember >> exactly) that for full dumps or for tapeless backup, holding disk should >> be set to no. When I set this to no, amanda complains that: >> [can't dump no-hold disk in degraded mode] >> >> Please tell me what this means. > >This is usually bad advice, because you loose all parallism of Amanda >and its ability to keep the tape streaming. The only exception i can >think of is if you are backing exactly one machine to the file output >driver, using no holding disk makes sense. > What you thought is exactly my config. I have one server running mail service, another for file sharing and is where amanda is. In the past, I had Tapeware installed on the second, which mounted the mail folder on the first, and backed up both. Recently, Tapeware has caused many crashes and I am forced to switch to amanda.Tapeware has the advantages of having a GUI, where I could schedule both full and incremental, specify which day to run the full, which to run inc. This makes the restoration task much easier b/c I knew which tape contained the last full, and started from it. If I have to rely on Amanda for choosing the full backup, is there any difficulty in locating it? Like looking for "tape" has the last full backup? My future plan is to immitate the old one: have Amanda run just on one machine, have it mount the mail machine using samba, and back both up just like a single one. This rather simplifies it because the mail machine has only one folder that needs to be backed up.
tapeless dump
I need to set up tapeless backup (onto a second hard drive), but it alway fails. I have tried the doc. from Clinton on Amanda website, but it does not work. Does anyone have a more detailed doc., or any suggestions to where I can find one?? Thanks,
[no subject]
I am trying to set up a tapeless backup, but keep receiving this error message in the mail FAILURE AND STRANGE DUMP SUMMARY: lnx200/Maillev 0 FAILED [can't switch to incremental dump] Anyone please tell me what this means and how to fix it. Thank you in advance.
