Hello,
This problem is surely related to the previous problems you were having with
that drive (or probably the IBM driver). The patch that you proposed was
very IBM specific, so probably not something that we would normally include
in the base code, and in looking at it I was not really convinced that it was
the definitive solution to the problem.
If you are getting empty files on your Volume it is very likely related to the
problem of spacing to the end of the volume data and/or forward spacing
files, which seemed to be cause of your previous problems. I still suspect
that there may be some driver problems related to SysV versus BSD tape
semantics. The Tape Testing chapter of the manual describes this is some
more detail.
You would probably do well to run the tape regression scripts tailored to your
drive. They will probably pull out a number of problems in the current
driver ... The regression scripts are documented in the Developer's guide.
Best regards,
Kern
On Monday 03 December 2007 14:05, Mariusz Czulada wrote:
> Hello again,
>
> I have another problem with bacula, related to finding EOD.
> Configuration: bacula 2.2.4 on CentOS 4.3, TS3310 (two LTO-3 drives)
> attached via FC.
>
> My pools:
>
> --------------------
>
> bacula=> select poolid, name from pool;
> poolid | name
> --------+-------------------
> 4 | Pl:Remote-file
> 5 | Pl:Remote-lib
> 1 | Pl:Localhost-tape
> 2 | Pl:REPfull-lib
> 3 | Pl:REPdata-lib
> 6 | Scratch
> (6 rows)
>
> ----------------------
>
> I defined (among others) four jobs to use "REP*" pools:
>
> 1. J:RepositoryFull-ncx
> 2. J:RepositoryFull-admin
> 3. J:RepositoryFull-access
>
> to use "Pl:REPfull-lib" pool, and
>
> 4. J:RepositoryFull-ncxdata
>
> to be stored on "Pl:REPdata-lib".
>
> Jobs 1, 2 and 3 run once a week on saturdays (1st - full, rest - diff).
> Job 4 runs daily (sun - full, rest - diff).
>
> Today I noticed, that volume used for "Pl:REPdata-lib" on previous week
> is marked as "error" and a new volume was taken from "Scratch". I didn't
> have debug output from SD, so all the rest is based on queries.
>
> ----------------------
>
> bacula=> select
> bacula-> mediaid, volumename, poolid, voljobs, volfiles, volblocks,
> volstatus
> bacula-> from
> bacula-> media;
> mediaid | volumename | poolid | voljobs | volfiles | volblocks | volstatus
> ---------+------------+--------+---------+----------+-----------+----------
>- 2 | file-0002 | 6 | 0 | 0 | 0 | Append 7 |
> A00020L3 | 6 | 0 | 0 | 0 | Append 8 | A00054L3
> | 6 | 0 | 0 | 0 | Append 9 | 000016L3 | 6
> | 0 | 0 | 0 | Append 10 | 000015L3 | 6 |
> 0 | 0 | 0 | Append 11 | 000017L3 | 6 | 0 |
> 0 | 0 | Append 12 | 000018L3 | 6 | 0 | 0 |
> 0 | Append 13 | 000019L3 | 6 | 0 | 0 | 0 |
> Append 14 | file-0003 | 6 | 0 | 0 | 0 | Append 4
> | A00024L3 | 2 | 6 | 80 | 1184664 | Append 5 |
> A00022L3 | 3 | 3 | 15 | 195403 | Error 3 | tape-0001
> | 1 | 5 | 5 | 4502 | Append 1 | file-0001 | 4
> | 19 | 1 | 77027 | Append 6 | A00051L3 | 3 | 3
> | 15 | 199745 | Append (14 rows)
>
> ---------------------
>
> Volume A00022L3 is marked with "Error", and volume A00051L3 is added to
> the pool 3 ("Pl:REPdata-lib"). Next, I checked jobs:
>
> ----------------------
>
> bacula=> select
> bacula-> jobid, name, starttime, endtime, poolid
> bacula-> from
> bacula-> job
> bacula-> where
> bacula-> name like 'J:Repository%';
> jobid | name | starttime | endtime
>
> | poolid
>
> -------+--------------------------+---------------------+------------------
>---+-------- 2 | J:RepositoryFull-ncx | 2007-11-29 11:35:33 | 2007-11-29
> 11:46:25
>
> | 2
>
> 3 | J:RepositoryFull-admin | 2007-11-29 11:49:33 | 2007-11-29
> 11:53:51
>
> | 2
>
> 6 | J:RepositoryFull-access | 2007-11-29 11:54:57 | 2007-11-29
> 11:57:21
>
> | 2
>
> 8 | J:RepositoryFull-ncxdata | 2007-11-29 12:02:28 | 2007-11-29
> 12:09:06
>
> | 3
>
> 21 | J:RepositoryFull-ncxdata | 2007-11-30 00:25:02 | 2007-11-30
> 00:30:18 | 3
> 34 | J:RepositoryFull-ncxdata | 2007-12-01 00:25:02 | 2007-12-01
> 00:25:06 | 3
> 36 | J:RepositoryFull-ncx | 2007-12-01 02:05:03 | 2007-12-01
> 02:24:53 | 2
> 37 | J:RepositoryFull-admin | 2007-12-01 02:25:01 | 2007-12-01
> 02:29:09 | 2
> 42 | J:RepositoryFull-access | 2007-12-01 02:45:01 | 2007-12-01
> 02:47:13 | 2
> 51 | J:RepositoryFull-ncxdata | 2007-12-02 00:25:02 | 2007-12-02
> 00:36:59 | 3
> 61 | J:RepositoryFull-ncxdata | 2007-12-03 00:25:02 | 2007-12-03
> 00:25:05 | 3
> 71 | J:RepositoryFull-ncxdata | 2007-12-03 10:27:06 | 2007-12-03
> 10:29:46 | 3
> (12 rows)
>
> ----------------------
>
> Jobs 8, 21 and 34 were stored correctly. Next, volume A00022L3 was
> unloaded, jobs 36, 37 and 42 were stored on A00051L3, also no problems.
> But then, for job 51 volume A00022L3 should be loaded and used, but for
> some reason it was not, and new A00051L3 was used.
>
> ---------------------
>
> bacula=> select
> bacula-> jobmediaid as jmid, jobid as jid, jm.mediaid as mid,
> m.volumename,
> bacula-> jm.firstindex as fi, jm.lastindex as li, jm.startfile as
> sf, jm.endfile as ef,
> bacula-> jm.startblock as sb, jm.endblock as eb, jm.volindex
> bacula-> from
> bacula-> jobmedia jm, media m
> bacula-> where
> bacula-> jm.mediaid=m.mediaid and
> bacula-> jobid in (select jobid from job where name like '%ncxdata%')
> bacula-> order by
> bacula-> 2, 1;
> jmid | jid | mid | volumename | fi | li | sf | ef | sb | eb |
> volindex
> ------+-----+-----+------------+-----+-----+----+----+----+-------+--------
>-- 45 | 8 | 5 | A00022L3 | 1 | 55 | 0 | 0 | 1 | 15499 | 1
> 46 | 8 | 5 | A00022L3 | 55 | 109 | 1 | 1 | 0 | 15499 | 2
> 47 | 8 | 5 | A00022L3 | 109 | 167 | 2 | 2 | 0 | 15499 | 3
> 48 | 8 | 5 | A00022L3 | 167 | 218 | 3 | 3 | 0 | 15499 | 4
> 49 | 8 | 5 | A00022L3 | 218 | 268 | 4 | 4 | 0 | 15499 | 5
> 50 | 8 | 5 | A00022L3 | 268 | 323 | 5 | 5 | 0 | 15499 | 6
> 51 | 8 | 5 | A00022L3 | 323 | 362 | 6 | 6 | 0 | 15499 | 7
> 52 | 8 | 5 | A00022L3 | 362 | 390 | 7 | 7 | 0 | 15499 | 8
> 53 | 8 | 5 | A00022L3 | 390 | 418 | 8 | 8 | 0 | 15499 | 9
> 54 | 8 | 5 | A00022L3 | 418 | 445 | 9 | 9 | 0 | 15499 | 10
> 55 | 8 | 5 | A00022L3 | 445 | 475 | 10 | 10 | 0 | 15499 | 11
> 56 | 8 | 5 | A00022L3 | 475 | 503 | 11 | 11 | 0 | 15499 | 12
> 57 | 8 | 5 | A00022L3 | 503 | 519 | 12 | 12 | 0 | 7914 | 13
> 61 | 21 | 5 | A00022L3 | 0 | 0 | 13 | 13 | 0 | 0 | 1
> 69 | 34 | 5 | A00022L3 | 1 | 5 | 14 | 14 | 0 | 1487 | 1
> 114 | 51 | 6 | A00051L3 | 1 | 54 | 0 | 0 | 1 | 15499 | 1
> 115 | 51 | 6 | A00051L3 | 54 | 111 | 1 | 1 | 0 | 15499 | 2
> 116 | 51 | 6 | A00051L3 | 111 | 167 | 2 | 2 | 0 | 15499 | 3
> 117 | 51 | 6 | A00051L3 | 167 | 220 | 3 | 3 | 0 | 15499 | 4
> 118 | 51 | 6 | A00051L3 | 220 | 267 | 4 | 4 | 0 | 15499 | 5
> 119 | 51 | 6 | A00051L3 | 267 | 322 | 5 | 5 | 0 | 15499 | 6
> 120 | 51 | 6 | A00051L3 | 322 | 364 | 6 | 6 | 0 | 15499 | 7
> 121 | 51 | 6 | A00051L3 | 364 | 391 | 7 | 7 | 0 | 15499 | 8
> 122 | 51 | 6 | A00051L3 | 391 | 419 | 8 | 8 | 0 | 15499 | 9
> 123 | 51 | 6 | A00051L3 | 419 | 447 | 9 | 9 | 0 | 15499 | 10
> 124 | 51 | 6 | A00051L3 | 447 | 475 | 10 | 10 | 0 | 15499 | 11
> 125 | 51 | 6 | A00051L3 | 475 | 503 | 11 | 11 | 0 | 15499 | 12
> 126 | 51 | 6 | A00051L3 | 503 | 525 | 12 | 12 | 0 | 10885 | 13
> 130 | 61 | 6 | A00051L3 | 1 | 5 | 13 | 13 | 0 | 1439 | 1
> 134 | 71 | 6 | A00051L3 | 1 | 5 | 14 | 14 | 0 | 1419 | 1
> (30 rows)
>
> ----------------
>
> On volume A00022L3 there should be 15 files. My experiences (or
> intuition?) told me, there can be a problem with number of files while
> seeking EOD with FSF, after tape reloading. I tried to verify it with
> bls. Here is the output:
>
> ----------------
>
> [EMAIL PROTECTED] sbin]# ./bls -V A00022L3 -v -j /dev/IBMtape1n
> bls: match.c:249 add_fname_to_include prefix=0 gzip=0 fname=/
> bls: butil.c:282 Using device: "/dev/IBMtape1n" for reading.
> 03-Dec 12:36 bls: Invalid slot=0 defined in catalog for Volume
> "A00022L3" on "Dev:TS3310-drv1" (/dev/IBMtape1n). Manual load my be
> required.
> 03-Dec 12:36 bls: 3301 Issuing autochanger "loaded? drive 1" command.
> 03-Dec 12:36 bls: 3302 Autochanger "loaded? drive 1", result is Slot 2.
> 03-Dec 12:37 bls: Ready to read from volume "A00022L3" on device
> "Dev:TS3310-drv1" (/dev/IBMtape1n).
>
> Volume Label:
> Id : Bacula 1.0 immortal
> VerNo : 11
> VolName : A00022L3
> PrevVolName :
> VolFile : 0
> LabelType : VOL_LABEL
> LabelSize : 177
> PoolName : Scratch
> MediaType : LTO-3
> PoolType : Backup
> HostName : pl-admbackup01
> Date label written: 29-Nov-2007 11:12
>
> Begin Job Session Record:
> JobId : 8
> VerNum : 11
> PoolName : Pl:REPdata-lib
> PoolType : Backup
> JobName : J:RepositoryFull-ncxdata
> ClientName : Fd:pl-admbackup01
> Job (unique name) : J:RepositoryFull-ncxdata.2007-11-29_12.02.26
> FileSet : FS:Repository-ncxdata
> JobType : B
> JobLevel : F
> Date written : 29-Nov-2007 12:04
> 03-Dec 12:38 bls: End of file 1 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:38 bls: End of file 2 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:38 bls: End of file 3 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:38 bls: End of file 4 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:38 bls: End of file 5 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:39 bls: End of file 6 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:39 bls: End of file 7 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:39 bls: End of file 8 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:39 bls: End of file 9 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:39 bls: End of file 10 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:40 bls: End of file 11 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:40 bls: End of file 12 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
>
> End Job Session Record:
> JobId : 8
> VerNum : 11
> PoolName : Pl:REPdata-lib
> PoolType : Backup
> JobName : J:RepositoryFull-ncxdata
> ClientName : Fd:pl-admbackup01
> Job (unique name) : J:RepositoryFull-ncxdata.2007-11-29_12.02.26
> FileSet : FS:Repository-ncxdata
> JobType : B
> JobLevel : F
> JobFiles : 519
> JobBytes : 12,500,453,179
> StartBlock : 0
> EndBlock : 7,913
> StartFile : 12
> EndFile : 12
> JobErrors : 0
> JobStatus : T
> Date written : 29-Nov-2007 12:09
> 03-Dec 12:40 bls: End of file 13 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:40 bls: End of Volume at file 13 on device "Dev:TS3310-drv1"
> (/dev/IBMtape1n), Volume "A00022L3"
> 03-Dec 12:40 bls: End of all volumes.
> End of physical tape.
>
> -------------------
>
> ????? End of file 13 ?????
>
> My idea is: job 21 stored *no* *data* in file 14, so it was assumed,
> that there is an EOD right after file 13 (last file of job 8) during
> seeking next files by bls, but also for job 51 after tape reloading.
>
> IS IT A BUG OR SOMETHING?
>
>
> ------------------
>
> Regards,
>
> Mariusz Czulada
>
> ----------------------------------------------------
> Płyta CD i kartka pocztowa "Pocztówka do Św.Mikołaja 2007" już
> w sprzedaży. Kupując tę płytę pomagasz dzieciom
> z domów dziecka. wwW.pocztowkadoswietegomikolaja.pl
> http://klik.wp.pl/?adr=http%3A%2F%2Fcorto.www.wp.pl%2Fas%2Fpocztowka.html&s
>id=122
>
>
>
> -------------------------------------------------------------------------
> SF.Net email is sponsored by: The Future of Linux Business White Paper
> from Novell. From the desktop to the data center, Linux is going
> mainstream. Let it simplify your IT future.
> http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
> _______________________________________________
> Bacula-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/bacula-devel
-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell. From the desktop to the data center, Linux is going
mainstream. Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
_______________________________________________
Bacula-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-devel
