On Fri, 24 Jul 2009, Kevin Keane wrote:
> I think even the cheapest flash memory has an MTBF of 10,000 write
> cycles, which means that an SSD used for spooling should last for 1000
> full backups - even more differential or incremental ones.
Wear levelling algorithms mean that deleted sectors ar
On Tue, 28 Jul 2009, Marc Cousin wrote:
> At peak during the night, we have around 40-50 write streams at the same time,
> and we are despooling to 3 LTO3 and 3 LTO1.
That's larger than my installation and I haven't got budget for a
dedicated array controller. We're using software striping on our
On Fri, 31 Jul 2009, Chris wrote:
> Hi
>
> I have a problem with tapes not being fully written
> Bacula reports them as "FULL" with less then 100GB written
How much less than 100Gb?
--
Let Crystal Reports handle the r
On Thu, 6 Aug 2009, Jeremy Koppel wrote:
> I thought that meant it wasn't going to actually do anything, but it did
> reduce the DB size to 6.5GB. I had actually stopped bacula before running it
> this time, so perhaps that had an effect. After that, I went ahead and ran
> dbcheck (thanks, Jo
On Wed, 12 Aug 2009, Ian Levesque wrote:
> I'm curious about using Bacula in a slightly different way than it was
> apparently intended. We have large directories of data that need to
> periodically be moved offline to tape. I've got a Scalar 100 with two
> LTO2 drives. I wouldn't be doing increme
On Wed, 12 Aug 2009, rorycl wrote:
> An important aspect of the system is that the tapes should be readable
> for 12 years, by other parties if necessary. From this point of view we
> like the idea of providing a CD with each tape set of the software
> needed to extract the contents
Judging from
On Thu, 13 Aug 2009, Ralf Gross wrote:
> I just had a bunch of 40 LTO-4 tapes that had problems during backups
> (or worse - only during the verify job afterwards). All from the same
> production date. So my trust in tapes is not that good anymore...
What do the SMART stats from those tapes look
On Thu, 13 Aug 2009, Dan Langille wrote:
> I think we have a problem:
I flagged this a couple of years ago with Overland changers.
> [dlangi...@backup01 bacula]$ sudo /usr/sbin/mtx -f /dev/sg4 load 3
> [dlangi...@backup01 bacula]$ sudo /usr/sbin/mtx -f /dev/sg4 status
> Storage Changer /dev/sg
People, PLEASE trim when quoting!
On Fri, 14 Aug 2009, John Drescher wrote:
> >> I can tell you that my IMPORT/EXPORT slot does work correctly after a
> >> slight modification of the mtx-changer script. Although after I put a
> >> tape in it the first time I run update slots bacula says there ar
On Fri, 14 Aug 2009, Dan Langille wrote:
> > I made a similar tweak to try and use the I/O slots and backed it out
> > for this reason.
>
> Can someone explain what IO slots are? And why I have 36 of them?
They are traditionally used for (un)loading individual tapes (mailslots)
without stopping
On Wed, 19 Aug 2009, Corey Shaw wrote:
> I ran the memtest on our bacula server last night. After 14 hours and 8
> passes it didn't find any problems. I'm at the end of my rope here. I'm
> trying a new virtual server to see if that fixes the issue.
I've found I can reliably trigger faults on a fe
On Wed, 9 Sep 2009, mehma sarja wrote:
> The user community should make an effort to offer pre-compiled fd's and sd's
> for various OSes. I was in the same dilemma.
The user community DOES make an effort to do this.
If they don't exist then noone running your OS has gotten around to it.
Why not
On Thu, 10 Sep 2009, Willians Vivanco wrote:
> Hi, i need to restore data of server A's backup on to server B
> filesystem... Any suggestion?
Just change the restore host and set the restore directory appropriately
during restore selection.
It works well, I do it fairly regularly.
--
On Mon, 14 Sep 2009, Silver Salonen wrote:
> Hello.
>
> I've got a problem in Bacula 3.0.2 - one backup of a server always fails with
> error:
>
> JobId 14638: Fatal error: sql_create.c:806 Fill Filename table Query failed:
> INSERT INTO Filename (Name) SELECT a.Name FROM (SELECT DISTINCT Name FRO
2.4.4 (not able to update yet)
I keep getting errors like this on update slots and incremental updates of
the bacula-dir server
Subject: Bacula: Admin Fatal Error of UpdateSlots.2009-09-13_12.30.43 Full
14-Sep 18:24 msslay-dir JobId 104670: Fatal error: sql_update.c:122 Update
failed: affected_
On Wed, 16 Sep 2009, Richard Mortimer wrote:
> The immediate failure is because the number of affected rows is zero. I'm not
> sure what the underlying cause could be (apart from the fact that JobId 104670
> is not in the Job table - pruning or a bug?)
More likely bug than pruning. It's being run
On Wed, 16 Sep 2009, Silver Salonen wrote:
> Yes, I did see this by googling, but I really doubted the server was out of
> disk space - there are several gigabytes free for MySQL DB and /tmp ain't full
> either. So I just couldn't believe it :)
As you've discovered: It's surprising how big the te
On Wed, 7 Oct 2009, Kael Fischer wrote:
> Hello,
>
> Reading the docs and experimenting has gotten me pretty far. I'm not
> having any problem backing up smallish jobs with spooling to disk and
> despooling to an LTO-4 drive in an overland arcvault12. But while
> backing up a 3TB RAID array, the
On Wed, 7 Oct 2009, Cedric Tefft wrote:
> The second phase (what you describe as DIA) is a result of using "Spool
> Attributes = yes" in your job defs. Changing this to no MIGHT solve
> your problem. OTOH, it might actually make things worse. I suggest you
> look up the Spool Attributes direct
On Fri, 2 Oct 2009, Klaus Troeger wrote:
> LTO-3 drive (equipped with "only" LTO-2 tapes, but )
>
> Physical drive performance is 28sec for 1 Gigabyte, so ~35MB/sec
>
> [r...@denbvsbcks1 disk1]# dd if=/dev/zero of=swapfile bs=1024
> count=100
> 100+0 records in
> 100+0 records out
On Fri, 6 Nov 2009, Cedric Tefft wrote:
> As a point of reference, I have a 70MB MyISAM database which, when
> dumped to a text file is about 80MB. That works out to about a 15%
> larger dump file than the native database. Your dump file is 2300%
> larger than the database (and counting). I do
On Mon, 9 Nov 2009, Alan Brown wrote:
> Nor can I. 20 billion rows dumped when there should only be only 400
> million in there and 3 billion ever written in.
>
> Arrgh...
Arrgh indeed.
To summarise: The file table was corrupt (innodb) and mysqldump was
jumping back about 50 mil
On Tue, 10 Nov 2009, Jesper Krogh wrote:
> Wonder if there is a feature request sitting in bacula for "accurate
> backups"? Looking at the projects file it doesn't look like it.
Yes - and it was implemented in V3, which is why it's not in the project
tree.
(Disclosure: We helped pay for it.)
On Wed, 11 Nov 2009, Graham Keeling wrote:
> ...and looking at the code now, it seems to me that what you want isn't
> possible without changing the code - or doing something like the following:
>
> bconsole << EOF
> add pool=poolname
> 1
> volumebasename
> 1
> EOF
or
echo add\\\n$NumVols\\\n$Vo
On Sun, 16 Dec 2007, Steve Thompson wrote:
> On Sun, 16 Dec 2007, David Legg wrote:
>
>> I'm sure this is all 'obvious' to the old hacks but is there a way to
>> prevent files being written into the mount point when no drive is
>> actually mounted?
>
> Just do your backups to a subdirectory on the
On Fri, 4 Jan 2008, Dan Langille wrote:
> I propose this new paragaraph at the top of the page:
>
> ###
> Yes, Bacula supports your tape drive. Bacula does not directly access
> hardware. If the tape drive is accessible from your operating system,
> then it should just work with Bacula. If you
On Thu, 10 Jan 2008, Mike Seda wrote:
> hi arno,
> i forgot to mention that i have:
> Maximum Concurrent Jobs = 2
>
> should i increase that number?
Yes - and read the fine manual, as the Maximum Concurrent Jobs directive
needs to be tweaked in several locations in several different config
file
On Thu, 10 Jan 2008, Jason A. Kates wrote:
> I wouldn't say that this is a bug. The shrink wrapped backup softare
> that you will pay 10's of K for work in the same way. If it's that
> much of an issue just do fulls for that file system.
Jason is correct, there are only a couple of (very expen
On Thu, 10 Jan 2008, pedro moreno wrote:
> My LTO-1 drive says 200GB compression 2:1, and the biggest full
> backup is < 179GB
That is pretty good going for a full backup.
Most of my full backups are about 198-210Gb per LTO2 tape (200Gb raw)
> I hope some day i can get more than 200GB.
A l
On Fri, 11 Jan 2008, John Drescher wrote:
> I am confused at that. I will have to try this the next time I get
> some new tapes. I did try the add command all the way to the question
> about the number of tapes to add. Since I do not have a single new
> tape I did not want to go any further...
If
On Thu, 10 Jan 2008, John Drescher wrote:
> Using label barcodes and then
> selecting drive 1 will allow this to continue while drive 0 is
> churning away.
Why do you use "label barcodes" at all?
I just add the tapes to appropriate pools. Bacula will automagically label
blank tapes the first ti
On Thu, 10 Jan 2008, Chris Hoogendyk wrote:
> For example, in the commercial realm, EMC Retrospect will properly
> restore incremental backups with deleted files removed. In the open
> source realm, Amanda will also properly handle deleted files in
> incremental backups.
When did Amanda gain that
On Fri, 11 Jan 2008, John Drescher wrote:
> How do I add blank tapes to a pool? I thought bacula just ignores them.
Inside bconsole, type "add" and follow the prompts.
Barcodes MUST be present and the required labels MUST match the barcodes.
AB
-
On Fri, 11 Jan 2008, John Drescher wrote:
> I only use label barcodes when I insert new tapes. I just don't have
> any new ones now..
Ok.
> My procedure is
> 1) Use a query command to determine which tapes in my archive are
> full. Of that I select which ones I want to remove.
> 2) Then I go to
On Wed, 16 Jan 2008, Ingo Jochim wrote:
> Hi Dan,
>
> a hard link looks like the original file. So it always should have the
> same date.
> Below you will see that the hard link (second line) has the same date at
> the "original" file.
> I created the hard link today but it also has the date from
On Thu, 17 Jan 2008, Kern Sibbald wrote:
> My best analysis at this point is that Bacula is writing
> to a tape volume and another job causes the tape to be unloaded, this causes
> a Bacula failure and also causes the Red Hat (CentOS) 5.1 kernel or driver to
> go belly up too (sometimes the dr
On Thu, 17 Jan 2008, Kern Sibbald wrote:
> Yes, in addition to being a Bacula bug, it is a Kernel bug, and I have filed a
> bug report, but I don't imagine they will fix it any time soon.
What are the reference details for the bug report?
On Thu, 17 Jan 2008, Kern Sibbald wrote:
>>> Yes, in addition to being a Bacula bug, it is a Kernel bug, and I have
>>> filed a bug report, but I don't imagine they will fix it any time soon.
>>
>> What are the reference details for the bug report?
>
> https://bugzilla.redhat.com/show_bug.cgi?id=4
On Fri, 1 Feb 2008, John Drescher wrote:
> Yes. You need to get the label barcodes working because you should
> never manually label tapes in an autochanger that has a barcode
> reader. If you label the tapes manually you are disabling the
> advantage of having a barcode reader to tell bacula what
On Tue, 5 Feb 2008, Tilman Schmidt wrote:
> I'm using a "Daily" pool for incremental backups on weekdays, and a
> "Weekly" pool for full backups on weekends. The "Weekly" pool has
> "Volume Use Duration = 8h" because the full backup goes into off-site
> storage.
Running full backups on a weekend,
On Tue, 5 Feb 2008, John Drescher wrote:
>> @echo off
>> xcopy /y /q /k C:\MySQL\data\bacula E:\backup\bacula-database\
>> exit
>>
> Isn't there a mysqldump for windows? I do not think this is a good way
> of backing up a database that is in use.
It isn't.
mysqlhotcopy can be used to copy out
On Fri, 8 Feb 2008, le dahut wrote:
> I'll use priority, since both jobs must be based on the same schedule
> included file.
If you use differing priorities, only one job will run at a time.
-
This SF.net email is sponsore
On Tue, 12 Feb 2008, le dahut wrote:
> Does this mean that the second job (the one with a higher priority number)
> will be "forgotten" ? or that it will be run once the first job has finished
> ?
It will be run after the first job is finished.
On Wed, 13 Feb 2008, Robin Blanchard wrote:
>> Although your status storage looks like a bug, have you tried to use
>> the mount command?
>
> How am I supposed to know which autochanger slot to mount, when it would
> seems bacula just randomly selects tapes from the pool (see attached
> screenshot
On Thu, 14 Feb 2008, Robin Blanchard wrote:
>>> How am I supposed to know which autochanger slot to mount, when it
>> would
>>> seems bacula just randomly selects tapes from the pool (see attached
>>> screenshot)?
>>
>> Mount slot 0
>>
>
> The restore job still has done nothing, resulting in queue
On Fri, 15 Feb 2008, Jesper Krogh wrote:
> > No need to delete the volumes from the catalog... just set their
> > status to something unusable... or, preferrably, disable them.
>
> Am I sure that bacula wont use a disabled "Full-run" as basis for a
> subsequent Incremental run?
It will, but just
On Thu, 14 Feb 2008, Robin Blanchard wrote:
> > Did you "update slots" after inserting the tape to the changer?
>
> There has been no change of tapes/slots in the library. The tape that is
> currently in /dev/nsa0 I assume is the tape last written to during a
> successful backup (which involved ro
On Fri, 22 Feb 2008, Rathinasamy, Bhaskaran (ext) wrote:
> The Autochanger company claims the following speed of writing.
>
> Data Transfer Rate (native): 80 MBps ( 22.8 GBph )
>
> Data Transfer Rate (compressed): 160 MBps ( 45.5 GBph )
>
> I feel the above is in Megabits/sec?
No, that's MegaByte
There is a Wikipedia page covering LTO specifications, tape speeds and
history. It should probably be in the Bacula FAQ and documentation.
http://en.wikipedia.org/wiki/Linear_Tape-Open
Note that the quoted native (uncompressed) speeds for LTO3 and LTO4 are
80MBytes/sec and 120MBytes/sec - this
On Mon, 6 Oct 2008, Keith Sudbury wrote:
> I am wondering what is the quickest way to blank / relabel a tape.
Use the "purge" command in bconsole
Bacula will recycle and relabel the tape itself, you don't need to do so
manually (and doing so will usually mess things up)
> I have a whole pile of
On Fri, 26 Sep 2008, John Drescher wrote:
> BTW, I would never use raid0 or LVM (without every PV being raided)
> for backup data that I cared about.
Spooled data isn't exactly worth keeping. After a bacula restart the
contents of those directories are useless anyway.
-
On Mon, 6 Oct 2008, Kjetil Torgrim Homme wrote:
>This directive is only implemented in version 2.5 and later. When
>set to {\bf yes} (default {\bf no}), this job may run even if lower
>priority jobs are already running. This means a high priority job
>will not have to wait for ot
On Thu, 2 Oct 2008, T. Horsnell wrote:
> Hi all,
> A couple of my volumes have accumulated "Error" status, and so are not
> being re-used. I've purged one of them and updated it's status to
> Append, but Bacula then seemed to mark it 'Full' and didnt use it. I
> then tried to re-label it with the
On Mon, 6 Oct 2008, Keith Sudbury wrote:
> > If you want to change the label, use the "relabel" command.
> >
> Can I change the job pool the tape uses as well?
Yes.
AB
-
This SF.Net email is sponsored by the Moblin Your Mo
On Mon, 6 Oct 2008, T. Horsnell wrote:
> Hm. I did just purge one of them at first, but bacula still didnt seem
> to want to use it.
It may not grab the tape straightaway. Bacula has its own algorithms for
tape reusa and a blank tape has higher priority than a purged one, etc.
> BTW, I had a bad
On Mon, 6 Oct 2008, John Drescher wrote:
> > Spooled data isn't exactly worth keeping. After a bacula restart the
> > contents of those directories are useless anyway.
> I believe the user was considering putting his disk volumes on a raid
> 0 ( or disk spanning lvm) because his raid5 write speed
On Mon, 6 Oct 2008, Kjetil Torgrim Homme wrote:
> > If you allow mixed priority jobs to run simultaneously, you also
> > need some way of flagging a job as "exclusive"
>
> this is accomplished by not setting Allow Mixed Priority on
> BackupCatalog.
OK, it wasn't clear in your earlier description
On Mon, 13 Oct 2008, Richerland Medeiros wrote:
> How do I use the default compression hardware native of LTO4?
> I have 1.4 TB of data and the tape is full when it comes to 900GB.
You are using the default compression already, else it would be full at
~795Gb
Compression only works on compressab
On Tue, 14 Oct 2008, Sebastian Lehmann wrote:
> > In general it's better to increase concurrent jobs and make sure
> > there's adequate spool space than switch this option on.
>
> oh, sorry, i didn't know that fact. Why is this not the recommended
> setting?
Because it means a single pool can mon
On Tue, 14 Oct 2008, Sebastian Lehmann wrote:
> Bacula will only use one drive per pool, so if you want to write to more
> then one drive you have to define several pools and direct the jobs to
> this pools.
Not quite correct.
That is the default setting, but Bacula can use several drives on one
On Thu, 16 Oct 2008, Arno Lehmann wrote:
> > I had this problem from the start. I ended up with a line in my
> > RunBefore script which disabled the backup job:
> >
> >echo "disable job=Job1" | /usr/local/bacula/bin/bconsole
> >
> > This disables any further scheduling of Job1 but leaves the c
On Fri, 17 Oct 2008, Dan Langille wrote:
> > It has sometimes happened to me also. Maybe it's because the tape is
> > positioned
> > different from where Bacula believes it is. Try to change the error
> > flag to
> > append and see if it will be accepted by Bacula. I have several
> > times been ab
2.4.2-1 (Scott Barninger's RPM, but that shouldn't make any difference)
Using an autochanger.
Restores which only use a single tape are fine.
For multiple tape restores, even thought he tapes are in various slots,
Bacula is insisting on loading all the tapes it wants from the first slot
it uses
On Mon, 20 Oct 2008, Martijn de Munnik wrote:
> Ok, is it a specific problem with Solaris or VXA2 or just tape drives in
> general? It seems as quiet a serious bug to me, every time I do a restore the
> data on my current tape is lost...
It's happening to me occasionally - linux_X86, FC robot and
On Mon, 20 Oct 2008, Dan Langille wrote:
> >> I believe this was a known bug in earlier versions of Bacula.
> >
> > It still happens.
>
> You have seen this in what version of Bacula?
Sorry, not on restore.
I'm getting tapes marked in error fairly regaulrly on an apparently random
basis. Right n
On Tue, 21 Oct 2008, Mark wrote:
> Hiya list,
>
> I added a bunch of volumes with the add command but both the volume
> sequence numbers and slot locations came out wrong.
Slot locations aren't a problem, just tell Bacula they aren't in the
changer and then run update slots afterwards.
> I then
On Mon, 27 Oct 2008, mehma sarja wrote:
> Quick math shows we need to backup 75%*(50 TB + 25 TB + 12.5 TB + 5TB) = 70
> TB. I can buy one 32 TB's RAID 5 array for about $12.7K not including any
> head ends. So, let's say I get by with 2 arrays - that's about $400/TB.
You assume you need to use d
On Wed, 5 Nov 2008, Jose Ildefonso Camargo Tolosa wrote:
> That's not the problem, it can listen on *all* interfaces (0.0.0.0), but
> then: how do you make "client X connect to A address, and client Y connect
> to B address"?
I do it by putting the specific address I want client Y to use in
/etc/
Bacula 2.4.1
I need to do a verify of the state of a filesystem at a particular date
vs what's in there now, as some files have been overwritten with nulls
whilst having their timestamps preserved (hardware problems).
I've tried using "Verify Disk to Catalog", but this only seems to say if a
fil
On Wed, 12 Nov 2008, Kelly, Brian wrote:
> Is it possible for me to pass params to bconsole to initiate a restore
> without having to deal with long lists of clients?
Yes.
The only parameter I've not been able to enter on the command line is
the date to restore from.
AB
--
On Thu, 13 Nov 2008, Ronald Buder wrote:
> > This is normal NFS behaviour - if a NFS server doesn't respond, the
> > processes accessing it wait in an uninterruptible state. They also do
> > not get notification of a problem by a signal.
> That's what I was afraid of...
It's a really _bad_ idea
On Thu, 13 Nov 2008, Hemant Shah wrote:
> > If at all possible, run a client on the NFS server to backup those
> > filesystems directly.
>
> It not possible to run bacula on deicated NAS server (e.g. NetApp
> Filer), so you have to back it up ove NFS mounts.
Some NAS servers are embedded windows
The subject says it all really.
I really don't want to convert all my match lines to
".*/[T|t][E|e][[M|m][P|p]/", etc etc
AB
-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest
On Fri, 14 Nov 2008, Kjetil Torgrim Homme wrote:
> Alan Brown <[EMAIL PROTECTED]> writes:
>
> > The subject says it all really.
> >
> > I really don't want to convert all my match lines to
> > ".*/[T|t][E|e][[M|m][P|p]/", etc etc
>
> &quo
I've been playing round with verify's DiskToCatalog and found that it
only compares file _attributes_, while I need to compare file hashes as
well.
(Background: we had some hardware problems with one of our RAID arrays
locking up under heavy load and while tranferring files from that to new
RAID
On Wed, 19 Nov 2008, Robert LeBlanc wrote:
> > You can write a udev rule to lock the drives down. I wrote one awhile
> > back to keep the changer device the same.
> >
> > dev6 ~ # cat /etc/udev/rules.d/55-bacula.rules
> > SUBSYSTEM=="scsi",ATTRS{vendor}=="EXABYTE*",ATTRS{type}=="8",
> > SYMLINK+="
On Wed, 19 Nov 2008, Martin Simmons wrote:
> > Can anyone assist on this?
>
> It should compare the signatures (aka file hashes) if configured to do so.
You'd think so, but all I get is "new" or "removed"
> What signature and verify options are you using for the fileset?
Here's an example
File
On Wed, 19 Nov 2008, Martin Simmons wrote:
> I think you need to add
>
> verify = 1
>
> to the options. The 1 means verify the SHA1, matching your signature line.
Yes, you're right, that was what was needed and it works now.
After rereading the "Using Bacula to Improve Computer Security" it m
On Thu, 20 Nov 2008, Cedric Devillers wrote:
> too) you can use /dev/tape/by-id/scsi-XX-nst that should be fixed by
> using specific id (instead of XX).
Looking at that directory it's only been created for the first tape drive.
The others have not been picked up.
> The only problem is t
On Thu, 20 Nov 2008, Robert LeBlanc wrote:
> Debian Lenny only picked up the changer in /dev/tape/by-id, but
> /dev/tape/by-path had both tape drives, but the names are horribly long:
> pci-:01:04.0-fc-0x500110a00058bd40:0x-nst-nst
> pci-:01:04.0-fc-0x500110a00058bd40:0x000
On Mon, 24 Nov 2008, Josh Fisher wrote:
> I ran into this with a couple of Vista laptops, and also with a
> nForce430 chipset XP machine. I have found it is not uncommon for
> Windows NIC drivers to be buggy.
Not just Windows.
We found that 3com Torando/Vortex/Boomerang family cards will start
b
Bacula 2.4.2, 64 bit, RHEL5
This is wierd, to say the least.
I have a subset of files in a directory I want to restore.
The directory doesn't show up in bconsole or in command line mode using
bgnome console but DOES does show up in the GUI-restore function on the
same bgnome console at the same
On Tue, 25 Nov 2008, Willians Vivanco wrote:
> Hi, i'm trying to make work Bacula in Debian/Lenny with a HP
> StorageWorks TapeLibrary MSL6000 and HP LTO3-Ultrium tapes of 800 Gb of
> capacity each one...
Nit: LTO3 is 400Gb capacity.
Any more than that is opportunistic based on compressability
terface.
What are the permissions of /dev/nst* and what userid are the programs
running as?
If these are correct:
What happens if you try to write to the tape using tar and related
programs?
>
> Regards
>
> Alan Brown wrote:
> > On Tue, 25 Nov 2008, Willians Vivanco wrote:
>
On Thu, 27 Nov 2008, Boris Kunstleben onOffice Software GmbH wrote:
> i am doing exactly that since last Thursday.
> I have about 1.6TB in Maildirs and an huge number of small files. I have to
> say it is awfull slow. Backing up a directory with about 190GB of Maildirs
> took "Elapsed time: 1 da
On Thu, 27 Nov 2008, Willians Vivanco wrote:
> > What are the permissions of /dev/nst* and what userid are the programs
> > running as?
> >
> The normal permissions root:tape
Bacula-sd and btape usually run as bacula:tape - as a test set the
/dev/nst devices world writeable and see if the error p
On Thu, 27 Nov 2008, Boris Kunstleben onOffice Software GmbH wrote:
> any idea if there is a better filesystem, im using ext3 on the clients
> and xfs on the director
I believe XFS copes fine with overstuffed directories.
Ext3 will perform a lot better if you use tune2fs and enable the following
Please reply to the list, not me personally. Other people may have the
problem now or in future and the archives are valuable resources.
On Thu, 27 Nov 2008, Willians Vivanco wrote:
> Willians Vivanco wrote:
> > Alan Brown wrote:
> > > On Thu, 27 Nov 2008, Will
On Thu, 27 Nov 2008, Kjetil Torgrim Homme wrote:
> this may be good for Maildir, but with Cyrus IMAPD, which uses short
> filenames like "1432.", "1433.", this *reduced* performance by a few
> percent in one test we did.
How many files/dorectory were you testing with? In some cases our users
were
On Thu, 27 Nov 2008, Tobias Bartel wrote:
> The database is a SQLite one, on the same machine but on a Software
> Raid 1.
SQLite is really oinly intended for testing, not production systems.
Switch to postgres or mysql, things should improve.
--
On Sat, 29 Nov 2008, Rodrigo Severo wrote:
> Is it safe to delete all Volumes with VolStatus = "Purged"?
Only if you do not plan to use them again.
Volumes which are Purged exist in the database, but have no other records
associated with them.
--
On Tue, 2 Dec 2008, Glen Davison wrote:
> The symptoms vary - sometimes there is an error at some point through the
> tape, and sometimes (suspiciously often) it seems to have lost the bacula
> label right at the start. These tapes were new, and the rate of these
> errors is far higher than I can
On Mon, 1 Dec 2008, Willians Vivanco wrote:
> I got this in dmesg output after running the btape test command:
>
> [ 1471.898470] st0: Sense Key : Aborted Command [current]
> [ 1471.898470] st0: Add. Sense: Information unit iuCRC error detected
> [ 1476.199199] st0: Sense Key : Aborted Command [c
On Sat, 29 Nov 2008, michael riehemann wrote:
> it´s great to have backups "forever"
Forever is relative. DVD R/RW and CD R/RW are both fragile media if not
kept under optimum conditions (cool dark place, a film refridgerator is
best)
AB
--
(Reposted with tweaked subject as noone has followed up)
Bacula 2.4.2, 64 bit, RHEL5
This is wierd, to say the least.
I have a subset of files in a directory I want to restore.
The directory doesn't show up in bconsole or in command line mode using
bgnome console but DOES does show up in the G
On Tue, 2 Dec 2008, Julien Cigar wrote:
> Yes the SCSI card is an Adaptec
Which model and revision?
Several older Adaptec scsi HBAs are _very_ sensitive to scsi bus
termination and length issues.
AB
-
This SF.Net email is
On Wed, 3 Dec 2008, Glen Davison wrote:
> I am pretty confident of cables and terminators. The card is more
> suspect, but I think the most likely culprit is the tape library itself -
> possibly the firmware on the drives. I've just had a protracted battle
> with this library, and a parallel ong
On Wed, 3 Dec 2008, Julien Cigar wrote:
> > Which model and revision?
>
> I posted the full output on the FreeBSD ML some times ago :
> http://lists.freebsd.org/pipermail/freebsd-scsi/2008-November/003706.html
Is that Ultra or LVD?
Ultra 160/320 single-ended scsi have maximum cable lengths in th
On Thu, 4 Dec 2008, Julien Cigar wrote:
> I think it's an Ultra, there are only two connectors on it (plus one to
> connect to the SCSI card). Actually it's configured as this :
> SCSI CARD == TAPE DRIVE == TERMINATOR
Is the scsi car internally terminated and is the terminator enabled?
T
On Fri, 5 Dec 2008, Alex Chekholko wrote:
> > - a built-in Fast Ethernet adapter (3com 3c509)
I have had a _lot_ of trouble with 3com Vortex/boomerang/tornado NICs
under high load - they tend to start emitting unswitchable packets which
splatter the entire network causing slowdowns on all machine
701 - 800 of 1196 matches
Mail list logo
