Richard Scobie wrote:
Bacula 5.0.2 Linux.
Summary:
Backup of Box A was done to Storage Daemon backup1
Update slots performed after tape removal
Tape was then moved to Storage Daemon flash
Fatal errors when attempting to restore files and requested tape mounted:
21-Oct 12:01 library1-dir JobId
Bacula 5.0.2 Linux.
I have a number of Jobs, written to LTO4 tape on a host, backup1, which
I am trying to restore from with the tapes loaded in a host called
flash. Director is running on host library1.
These restores are failing with either an error:
RestoreFiles.2010-10-21_13.54.01_08
Dan Langille wrote:
I don't understand here why it attempts to use Drive-1, which is the
drive on the creation host, backup1.
My guess: because that's where the Catalog says is it?
A list media for the Archive pool shows the In changer and Slot
flag for AR0030L4 to be 0.
Regards,
Richard
Dan Langille wrote:
Instead, what you're saying is:
- backup of Box A was done to Storage Daemon backup1
- tape was then moved to Storage Daemon flash
Yes, sorry for the confusion.
Does the Director know that the tape has moved from one system to another?
Only in the sense that once the
David Noriega wrote:
both drives and 2) has as much data as these drives can take pouring
into them, since this is taking too long to do a backup. I've read a
One solution to feed your drives at full speed, is to build a low cost
server containing an array of SATA drives, sufficient to hold
David Noriega wrote:
Its an idea, but not what I asked. I would have gone with a completely
disked based backup, but that kind of hardware is expensive to do it
right. I asked about multiplexing. I figure I could take /home and
My apologies, I distinctly thought you said as much data as these
Dan Langille wrote:
Surely if a tape is mounted in a drive and the label is read, this
should become the new location for this tape?
Are you issuing a mount command after inserting the tape into the other drive?
Yes and when the restore job is submitted, here are the results.
Restore job,
Richard Scobie wrote:
Device status:
Device FileStorage (/mnt/exp1/bacula) is not open.
Device flash_LTO-4 (/dev/nst0) is mounted with:
Volume: AR0030L4
Pool:*unknown*
Media type: LTO-4
Total Bytes Read=2,048 Blocks Read=2 Bytes/block=1,024
Unless I have missed something, it appears that if you have an
autochanger full of tapes and it becomes unavailable for whatever reason
- say the PSU fails, there is no way of accessing the tapes in another
drive, as the system thinks are still loaded in the library and will not
allow a
I have a tape I am trying to restore from on a local LTO drive.
It was created at a remote location, in a library which is no longer
attached to the SD there and the update slots command was not done
after this tape was removed, so the Director still sees the tape as
residing in the remote
Jeremiah D. Jester wrote:
I'm confused. The tapeinfo output for /dev/sg1 says it is a 'Tape Drive' and
of type 'TD4' or LT04, the new supported capacity. Sg0 is the hard drive. I
get no output for devices sg3-sg10. Please explain.
What is the output of:
cat /proc/scsi/sg/device_strs
On my
Paul Mather wrote:
By definition, random data are not compressible. It's my understanding that
the compressed capacity of tapes is based explicitly on an expected 2:1
compression ratio for source data (and this is usually cited somewhere in the
small print). That is a reasonable estimate
I have just added 10 unlabeled, barcoded tape to a library that contains
10 labeled, barcoded tapes.
From the bat application, I have issued the following commands:
update slots
label barcodes
At this point the bat console shows a list of the tapes that will be
labelled, but this includes
John Drescher wrote:
Also, in this scenario, are the tapes labeled in ascending slot order or
ascending barcode order?
The latter.
Actually it is labeling them in slot order which is annoying. In future
I will remember to load all the tapes with ascendinging barcode numbers
in ascending
I have a weekly backup set which has a retention time of 13 days.
I have just been asked to keep a set that was written 12 days ago.
Can someone please tell me how I can extend the retention time for this job?
Regards,
Richard
skipunk wrote:
After returning to my office this morning, I had found the server was
connected to a 10/100 switch. I had moved it to the rack and connected it to
the switch 10/100/1000 and nic 1 pulls 1000 on ethtools
nic 2 pulls 100 on ethtools.
After futher investigation, it looks like
skipunk wrote:
What would it take to clear up the network bottleneck. I was looking at
another server that came with our gaming system. Same server (memory, hd,
etc) and tape library, running on win2k3 and netbackup with no issues. The
only difference is, it's running 4 nic's. So it's
skipunk wrote:
Status update. I just finished the first backup job which was 3.6 Gb. It
took 2.5 hours to run. avg speeds finished around 500k/s.
I've gone as far again to reboot the library and the server and still no
changes. I'm not sure what i'm over looking. I have 1T of data to
Simone Martina wrote:
Someone, from this ML, has bought a new tape library working with Linux
and mtx-changer?
While not on the scale you require, I am having very good results with
an HP MSL2024 with a single LTO-4 1840 drive fitted. I imagine the
larger family members will be no different.
Gary R. Schmidt wrote:
I think I'll hold off updating for a bit longer. Sigh.
Another bat bug I have just discovered going from 3.0.3 to 5.0.2, is
mounting tapes in an autochanger from the command window:
bat output in console pane:
mount
Select Storage resource (1-3): Automatically
Gary R. Schmidt wrote:
I think I'll hold off updating for a bit longer. Sigh.
I have just discovered the issue of bat being unusably slow restoring on
5.0.X is known about by the developers and is being worked on.
Regards,
Richard
Joseph Spenner wrote:
Ok, I just used bconsole and recovered a directory in little time at all.
So, bat is the problem. I can use bconsole instead. I might try to go back
to using 1TB files and see how it performs as well.
Thanks for the tip!
I can confirm this and will post more fully
I installed 3.0.3 a month or so prior to 5.0.0 being released, with a
view to using bat for restores, as it made life much easier for
commandline-phobes, and it performed very well.
It was then updated to 5.0.1 and restores of anything more than a
handfull of files became unusably slow. I
Bacula 5.0.1 and an HP MSL2024 changer.
The drive is empty and the tape I wish to restore from is resident in
the library.
After choosing some files to restore, I get the following:
restore jobid=163 client=backup1-fd_arc file=?restore_476707713_20
done
3 files selected to be restored.
Craig Ringer wrote:
I'm interested in ext3, ext4 and xfs. I should probably look at zfs too,
but don't have any hosts that it runs on usefully and don't really have
any personal interest in it.
You may find the XFS mount directive, filestreams of benefit here.
There is not much documentation
Gary R. Schmidt wrote:
And using the 64-bit XFS will also better[1] than the standard 32-bit XFS.
That would be using the inode64 mount option.
Regards,
Richard
--
Download Intel#174; Parallel Studio Eval
Try the new
Craig Ringer wrote:
Snip interesting crypto discussion...
I'd use the hardware encryption (which presumably has no performance
impact), that is an option on this autochanger, except they want $2500
for it...
Probably because it has a custom ASIC for the crypto algorithm in use to
allow it
I have a 2.8GHz Core i7 machine backing up uncompressable data spooled
onto an 8 drive RAID5, to LTO-4 tape.
Our requirements now dictate that data encryption must be used on the
tapes and having configured this, it seems that one core is saturated
encrypting the data and the result is that
John Drescher wrote:
I think you can update slots with the single drive running jobs. You
can not label barcodes however.
I can confirm this, as I did this a couple of days ago, although watch
out for the size of the emailed job report, which for me was just under
4MB and contained thousands
Bob Cousins wrote:
- Tape verification option -- verify that the bits read back from the
tape are the same as those originally written. Spooling to disk makes
this easier. Also, if a tape fails verification, it should be easy to
restart the write/verify cycle with new media.
I used to do
Is there an extra command that needs to be run, in order to free volume
space after a job is cancelled?
A job was started on an appendable LTO4 tape and for various reasons it
was decided to cancel it after it reached the end and was requesting a
new volume to be loaded.
After the
Timo Neuvonen wrote:
Cancelling a job does not free the storage that the job occupied before it
was cancelled.
Thanks. I guess there is a reason for this but it seems counter
intuitive - I'd expect if something is cancelled, for all resources to
be freed.
Regards,
Richard
Alan Brown wrote:
As Timo said, in a tape, jobs may be interleaved. Because of that it's
dangerous to automatically purge one because of a job failure.
That makes sense.
Note that if you delete the failed JobId from the database and that job
is the only one on the tape, it will free up for
I am just testing bacula prior to deployment and restored 7 files from
one directory from a backup of 3TB.
This backup is stored on 3 LTO4 volumes and in order to restore these 7
files, it read from the first and third LTO4 volumes, when I would have
expected them to all be adjacent to one
Phil Stracchino wrote:
If that's a backup of a single host, that sounds ... bizarre. The only
explanation I can think of is that there's a directory tree sitting in
somewhere between those seven files in the directory, that contains a
vast quantity of data. If you look at your filesystem,
Can someone please confirm or not, whether Catalog entries are
duplicated when a job of type Copy is run, or do both the original and
the copy share the same catalog data?
Regards,
Richard
--
SOLARIS 10 is the OS for
Bob Hetzel wrote:
My next thought was to try use tar to read and write from a tape. That
works fine: I've been able to back up an entire directory and restore
files from it. I've also back up a single file and restored that file,
comparing it to the original file and no differences were
Bob Hetzel wrote:
The auto test passed, but I now cannot mount any tape from within
bacula, either completely new (i.e. unlabeled) or filled completely with
bacula backups. It always thinks the tape is not labeled, giving errors
like these:
This is exactly what I was seeing and the problem
Robert Hartzell wrote:
I know everything can be manually changed... Just trying to get a clear
understanding of what exactly happens to a volumes attributes as bacula
moves it in and out of the scratch pool.
After filing the observed behaviour of a scratch volume not inheriting
the Recycle
Robert Hartzell wrote:
What I didnt expect to happen is that if a volume is moved from the
scratch pool to another pool and then update volume from pool is run the
RecyclePool attribute is set to *None*. That seems to contradict the
above explanation unless that is what is meant by modify it
Bob Hetzel wrote:
Has anybody gotten 5.0.0 to label tapes properly? I upgraded to 5.0.0 a
couple days ago and backups worked that first night until they ran out of
appendable (i.e. already partially written) media. Since that point, I've
been completely stuck.
Is bacula the only thing you
ganiuszka wrote:
In Client Run After Job directive %v will not be replaced because it
can be replaced only on Director side.
So if you has:
Client Run After Job = /path/script.sh %i %v
than
%v will be empty.
OK thanks. I actually have the %i_%v inside the shell script, so along
with
Robert Hartzell wrote:
It would seem reasonable (to me anyways) that the attributes would be
updated to reflect the current pool. There isn't much in the way of
documentation regarding this feature so hopefully someone can verify one
way or the other and I dont have to set up a test to find
Robert Hartzell wrote:
hmmm... looks like I need to do some testing. Do you have any volumes
that have been recycled and moved back to the scratch pool?
Yes. A tape that was initially labeled in the Full pool has been
recycled back into Scratch and is showing Scratch as its recycle pool.
I
Richard Scobie wrote:
Robert Hartzell wrote:
hmmm... looks like I need to do some testing. Do you have any volumes
that have been recycled and moved back to the scratch pool?
Yes. A tape that was initially labeled in the Full pool has been
recycled back into Scratch and is showing Scratch
Robert Hartzell wrote:
Which says to me that RecyclePool should be defined in the scratch pool
and not the Full pool.
I guess this would work if you never label media as anything other than
Scratch.
Does this mean that the RecyclePool attribute is
attached to the volume when it's initially
I have a Client Run After Job which runs a bash script which ends up
writing a file on the client, %i_%v.txt
It all works except the substitution is not occuring and it is literally
writing the file %i_%v.txt.
Is this expected?
Regards,
Richard
John Drescher wrote:
I believe other users mentioned that restarting bat fixed the issue.
And the issue only occurred on the first run of bat.
I have to date been unable to see the console on one bat install and
have a bug open:
http://bugs.bacula.org/view.php?id=1468
(username/password
Marc Schiffbauer wrote:
Yes, I saw this here, too. Then on another startup of bat, it worked
properly.
So now, If that happens, I close bat and start it again until the
console is there...
Using QT4.6.1 here.
Still looking into it here.
One machine with qt-4.5.3-9.fc11.i586 never shows
Have just updated all components to 5.0.0 and a remote bat connection
does not display the console window - it just stays on the last selected
page.
When starting bat, it just goes up with a blank page, but all other
pages display correctly.
Anyone else seeing this?
Regards,
Richard
I note that it is possible to run clone jobs by using Run = jobname of
current job in the Job resource.
What I would like to do is make 2 tapes (Volumes) that are exact copies,
but with only one catalog entry, used if either tape is restored.
1. Does cloning acheive this?
2. Can cloning be
Daniel wrote:
Hello,
I am having a problem where all jobs but the first immediately after starting
bacula-dir hang waiting for execution.. I have no further information on
why they are waiting, nor what they are waiting on. They have waited nearly
a week. All jobs I queue after them
Dan Langille wrote:
ganiuszka wrote:
Here is a instruction about Dell PowerVault 124T and LUNs under Linux.
http://support.dell.com/support/edocs/stor-sys/124t/en/124tvs16/install.html#enabling_lun_linux
Subject Enabling LUN Support in Linux
Maybe this is the solution.
This is what I see
Dan Langille wrote:
Fusion MPT SPI Host driver 3.12.19.00rh
ACPI: PCI Interrupt :11:08.0[A] - GSI 16 (level, low) - IRQ 169
mptbase: Initiating ioc0 bringup
ioc0: 53C1030: Capabilities={Initiator,Target}
scsi2 : ioc0: LSI53C1030, FwRev=01032700h, Ports=1, MaxQ=255, IRQ=169
Vendor: IBM
Dan Langille wrote:
Dan Langille wrote:
I'm helping a client with a Linux system. I see the tape drive, but
cannot find the tape changer. Clues please.
From dmesg:
Also:
Linux version 2.6.9-78.0.8.plus.c4smp (mockbu...@builder10.centos.org)
(gcc version 3.4.6 20060404 (Red Hat
Richard Scobie wrote:
Thanks for the followup, but after finding the thread referenced below
yesterday, I switch the SCSI HBA and library into another box (different
motherboard) and it is humming along fine using the same config files.
Granted this box is running Fedora 8, not Fedora 11
Brian Debelius wrote:
Humor me if you will. Go back to default. Make sure your tape drive is
set to variable blocks using mt -f /dev/nst0 setblk 0, then load a tape,
and use tapeinfo -f /dev/nst0 to verify it is set to zero. Now comment
out the Maximum Block Size from the sd. Test it with
John Drescher wrote:
Have you tried to stop bacula-sd and then execute the mtx-changer
script directly from the shell? You will probably have to look at the
code for the parameters.
Hi John,
Yes, I have successfully been able to do this a number of times and also
run through the btape
Brian Debelius wrote:
Try turning on debug in mtx-changer. Modify the wait_for_drive()
function to add another debug statement to confirm that the device is
ready, and that the function did not just time out.
wait_for_drive() {
i=0
while [ $i -le 300 ]; do # Wait max 300 seconds
if mt -f
ext-daniel.haw...@nokia.com wrote:
Hi Richard,
Anyone else running this library I can compare notes with?
Not anymore :) Used to use an SAS LTO3 based MSL2024 on a project I worked
at a while back. TBH it *just worked* and worked well too. Would recommend
one to anyone that needed an
I have an MSL2024 library I have been battling to get running for the
last couple of days and am about out of ideas. It is the parallel SCSI
attached LTO-4 version, connected to an HP LSI based HBA.
I have bacula configured and it passes the mtx-changer test commands
recommended in the manual
You probably need to add a wait in the mtx-changer script just after
the load. This wait will make sure the tape is in the drive and has
completed the loading process. For some systems the script does not
wait long enough for the tape drive to finish.
John
Thanks John. I already have a 30
Richard Scobie wrote:
You probably need to add a wait in the mtx-changer script just after
the load. This wait will make sure the tape is in the drive and has
completed the loading process. For some systems the script does not
wait long enough for the tape drive to finish.
John
Thanks
Brian Debelius wrote:
That would be 16M. Isn't the SD hard limited to 1M?
The maximun size-in-bytes possible is 2,000,000
From the SD Configuration Maximum block size directive.
Regards,
Richard
--
This SF.Net
Brian Debelius wrote:
Shamless bump. Does anyone have any insight into this?
Thanks
brian-
The difference between the dd tests and bacula backups is that catalog
writes are occuring in the latter.
Perhaps you have a database bottleneck? Is the database stored on the
RAID also?
Regards,
Bob Hetzel wrote:
1) Bacula is indeed able to seek within a tape. If you're having trouble
with this functionality you need to look at the storage daemon options at
Thanks Bob, I had misinterpreted something I read earlier.
2) If you pick a file that's on one of the other 9 tapes in your
If a tape is damaged in a multi volume archive, is it possible to
recover all the data from the rest of the set?
I understand Bacula is not currently able to seek within a tape in order
to recover a file - the tape must be read sequentially.
In a multi volume archive, does it need to read
67 matches
Mail list logo
