For a long time now, we have been applying the following patch to Amanda: diff -u -r1.1.1.3 -r1.4 ... +++ client-src/sendsize.c 2000/11/30 21:09:12 1.4 @@...
2319
Marc W. Mengel
mengel@...
Dec 1, 2000 4:21 pm
... It is: <bldsunos28> uname -a SunOS bldsunos28 5.8 Generic_108528-02 sun4u sparc SUNW,Ultra-5_10 <bldsunos28> /usr/bin/perl -v This is perl, version...
2320
Patrick J. LoPresti
patl@...
Dec 1, 2000 5:22 pm
After doing "su -m amanda", I tried to manually launch amdump. I was promptly told that I had to run amdump as the "amanda" user. I suspect that amdump is...
2321
John R. Jackson
jrj@...
Dec 1, 2000 6:06 pm
... I suppose we could do that, but I'm not sure host names follow patterns as much as disk names do, so the gain might be less. ... You probably could, it...
2322
Jean-Louis Martineau
martinea@...
Dec 1, 2000 6:40 pm
... All machine in my cluster are named clust00 to clust16 ... Both host and disk will match by word with either (.) or (/) as word delimiter and each word is...
2323
John R. Jackson
jrj@...
Dec 1, 2000 7:03 pm
... I see no difference in the above and my proposal. Mine will allow any backup image (stream) to be a candidate for writing to tape. It does not have to be...
2324
John R. Jackson
jrj@...
Dec 1, 2000 7:18 pm
... Mine are champion, lookout, fortress ... :-). ... So you're going to change all the RE's to glob's, e.g. amrestore and amadmin? And you'll update the docs?...
2325
Tobias Weingartner
weingart@...
Dec 1, 2000 10:08 pm
... Yes, there is a difference. In your scheme, reliability (through extra tape marks) is directly related to the performance each stream that you are writing...
2326
John R. Jackson
jrj@...
Dec 2, 2000 12:45 am
... You didn't. I must be getting really tired of this week, based on the tone of the letters I've been posting lately :-). ... I agree the number is a...
2327
John R. Jackson
jrj@...
Dec 2, 2000 1:45 am
This is from a month ago. The original question was about amdump not working because it claimed it was being run by the wrong user. That ... And then someone...
2328
John R. Jackson
jrj@...
Dec 2, 2000 1:57 am
... There's no such thing as (difference between) real and effective UID unless you're dealing with setuid programs, which amdump is not. ... The change was...
2329
Alexandre Oliva
oliva@...
Dec 2, 2000 8:41 am
... C- write a disklist file with the selected disks and give it as an additional argument to amreport. -- Alexandre Oliva Enjoy Guarana', see...
2330
Alexandre Oliva
oliva@...
Dec 2, 2000 8:43 am
... Actually, it's the FILTER API, that follows a similar principle. -- Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/ Red Hat GCC...
2331
John R. Jackson
jrj@...
Dec 2, 2000 8:02 pm
For those following my saga of the file system too big to fit on tape, it did turn out to be a trivial (attached) patch to amrestore to make it alter the...
2332
Mitch Collinsworth
mitch@...
Dec 2, 2000 8:23 pm
... I'd be interested in knowing what difference 'fast' makes, both in time and in size. ... Not bad... just unpredictable. ... My experience was that 'fast'...
2333
David Wolfskill
dhw@...
Dec 2, 2000 9:13 pm
I finally made the time to fetch 2.4.2 (after using a slightly-hacked 19991216-beta1 for quite a while). I used the same configure invocation, then proceeded...
2334
Dietmar Goldbeck
dietmar.goldbeck@...
Dec 2, 2000 9:30 pm
... Always using best compression is bad for the tape drive wear and restore times. I sometimes use "amrestore -c /dev/nst0 host" and since my images mostly...
2335
John R. Jackson
jrj@...
Dec 2, 2000 9:59 pm
... Why would it affect tape drive wear? And restore time (uncompress) is not particularly affected by the level of compression, or so I thought. ... The -c...
2336
Dietmar Goldbeck
dietmar.goldbeck@...
Dec 3, 2000 12:19 am
... If you have a tape with mostly compressed images and one huge image with no compression and run amrestore -c ... this will pull of data _very_ _slow_ from...
2337
Patrick J. LoPresti
patl@...
Dec 3, 2000 2:15 am
... Thank you for the patch, but the whole algorithm still strikes me as bogus. $LOGNAME is just meaningless; it has no necessary relation to permissions....
2338
John R. Jackson
jrj@...
Dec 3, 2000 2:43 am
... It's only used as a last resort. But I agree with you. ... For one thing, this would suffer from the same problem you mentioned about whoami. So either...
2339
Patrick J. LoPresti
patl@...
Dec 3, 2000 2:49 am
... Not at all! "whoami" maps a UID to a name, but there can be multiple names for a single UID, so it might return the wrong name. getpwnam() maps a name...
2340
John R. Jackson
jrj@...
Dec 3, 2000 5:08 pm
After thinking on it a bit more, you might as well hit these, too: server-src/dumper.c server-src/taper.c Also, while I was hunting around, there is already...
2341
Jean-Louis Martineau
martinea@...
Dec 3, 2000 5:45 pm
... D- Add a new option to amreport where it will ignore the missing result. amreport report "not in disklist" for disk in the log that are not in the...
2342
Jean-Louis Martineau
martinea@...
Dec 3, 2000 5:56 pm
... yes. ... It will work, I add a separators at the begin and end of the entry in the disklist: sda -> /sda/ /dev/sda -> /dev/sda/ /...
2343
Jean-Louis Martineau
martinea@...
Dec 3, 2000 5:59 pm
... amrestore should be able to chunk the image. Jean-Louis -- Jean-Louis Martineau email: martineau@... Departement IRO, Universite...
2344
John R. Jackson
jrj@...
Dec 3, 2000 6:14 pm
... Handle them, yes. Correctly, no. There are valid cases where it should report missing results. We must not get rid of that. ... I thought that meant you...
2345
John R. Jackson
jrj@...
Dec 3, 2000 6:33 pm
... Yeah!!! :-) ... OK, it does look like '*' will match everything. I didn't realize (had sort of forgotten) you were going to allow me to just enter...
2346
John R. Jackson
jrj@...
Dec 3, 2000 6:35 pm
... That may be, but it has nothing to do with fixing the header bug and is **way** more than I want to tackle for right now :-). I assume, since amrestore...
2347
Jean-Louis Martineau
martinea@...
Dec 3, 2000 6:53 pm
... That's for your TODO. ... 2 GB file size limit. ... That's what I had in mind. The chunker is a separate process on 2.5.0 but it exchange many command...
