Koozali.org: home of the SME Server

[Announce] Dar - Disk Archive (alpha build)

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #30 on: September 19, 2006, 07:07:23 PM »
byte---
It's still running after 'kill 14219'.
Have you got any 'Parazone/DDT'-flavoured commands?
ie anything stronger?
----best wishes, Robert

Offline byte

  • *
  • 2,183
  • +2/-0
[Announce] Dar - Disk Archive (alpha build)
« Reply #31 on: September 19, 2006, 07:11:23 PM »
Then you will need to have a read and find out how to use the "kill" command more effective.
--[byte]--

Have you filled in a Bug Report over @ http://bugs.contribs.org ? Please don't wait to be told this way you help us to help you/others - Thanks!

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #32 on: September 19, 2006, 07:27:29 PM »
byte----
I see nothing in man kill(1) to strengthen the scope of 'kill 14219'.
It mentions kill(2) but I can't for the moment see how to access (2).
Oh well
#console
#reboot
should do the trick;~)
----best wishes, Robert

Offline byte

  • *
  • 2,183
  • +2/-0
[Announce] Dar - Disk Archive (alpha build)
« Reply #33 on: September 19, 2006, 07:33:18 PM »
Going OT now, but you would have used kill -9 pid number
--[byte]--

Have you filled in a Bug Report over @ http://bugs.contribs.org ? Please don't wait to be told this way you help us to help you/others - Thanks!

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #34 on: September 19, 2006, 07:39:39 PM »
byte----
Actually I tried 'kill (9) 14219' but I should've thought of '-9';~)
Needed to reboot to eyeball the bootup graphics for EL or ELsmp.
Will continue in other thread (more point).
----best wishes, Robert

Offline jpl

  • *
  • 112
  • +0/-0
[Announce] Dar - Disk Archive (alpha build)
« Reply #35 on: September 19, 2006, 08:25:02 PM »
Quote from: "icpix"
jpl----
It still doesn't make sense to me.
Yes, I know I am not a programmer but I can do maths.
Your code says:
$timeout = ($timeout*3600)-60;
...that '-' minus symbol... maybe it should be a divide symbol '/'?
Try for 1hrs when I return to the control panel it is 3540 (1x3600)-60.
Try for 2hrs when I return to the control panel it is 7140 (2x3600)-60.
----best wishes, Robert


The meaning of the formule is : n hours minus 1 minute

The minute is kept in case of 24h to cleanly stop running backup before cron start a new backup job.

About the panel : why do you come back to the configuration panel ?

The given modification is just a temporary solution. Yes if you enter 1 hour and go again to configuration you see 3540, and must re-enter 1.
But it permits you to do configuration. All will be ok in next rpm.

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #36 on: September 19, 2006, 08:47:55 PM »
jpl----
Understood, my thanks.
I look forward to the new rpm.
Had difficulty stopping the DAR process when I could see it was
'doing' nothing at all productive except using nearly 98% CPU.
Is this an appropriate way to stop it...
root# kill -9 PIDnumber
...where the PIDnumber is the PID shown for DAR in 'htop'?
----best wishes, Robert

Offline jpl

  • *
  • 112
  • +0/-0
[Announce] Dar - Disk Archive (alpha build)
« Reply #37 on: September 19, 2006, 11:09:19 PM »
Quote from: "icpix"
jpl----
Had difficulty stopping the DAR process when I could see it was
'doing' nothing at all productive except using nearly 98% CPU.
Is this an appropriate way to stop it...
root# kill -9 PIDnumber


Yes it'is a good way to kill it, but don't forget to kill also the mount process for the share if it runs.

For 95% or 98% CPU it is also normal.
Nigthly backup often occurs when no other task needs CPU. Also by default dar-workstation backup uses maximum compression algorithm and... maximum CPU disponibility. But compression effort can be decreased from 9 to 0 by templating. In this case load of the CPU must not be very hard (but I can't think it's a good idea to backup while your server is busy)

Perhaps (apart from the timeout format bug) all your problem is one unique problem : user used in configuration doesn't have rights to write on the mounted share.

In this case maybe the script detect that the share is correctly mounted, but it doesn't detect the fact that you cannot write in it. Could you - after a reboot - start a backup and see if the shared folder is really mounted and if the server can write in it ?

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #38 on: September 19, 2006, 11:23:48 PM »
jpl----

SME7 box has dual 3.4GHz Xeons. At the moment it serves just me.
Later (when more problems are fixed) it goes on-line for web sites
and email but still used (on gigabit intranet) with picture library for me.
98% CPU utilisation for one CPU is OK. My problem was/is that I now
realise SME7 box now uses only one CPU and not two CPUs;~( That
was why I needed to kill the DAR process! Something is now different
between SME7 beta's and gold. What - Why - How to fix! Reload...?

Everywhere here user is admin/root. Not best I know but only me.

Yes, I try again, now or soon. For some reason I am very good
at breaking new software/hardware - my curse? ability? whatever;~)

----best wishes, Robert

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #39 on: September 20, 2006, 01:16:37 AM »
jpl----

While testing or at least trying to use/run DAR was dealing with a CPU
issue that I have now been able to fix with everyone's help. Don't think
that it affected DAR except the SME7 box not having its usual 'grunt'.

Trying to implement another test run. It's now just after midnight.
There may be an issue to address... it's between the control panel
checking algorithm and the use of your AM/PM and whatever DAR
is expecting. Can't seem to find the right combination to kick it off...
0.01 AM
12.01 AM
...around this particular witching hour. To me this time is 0001hrs
(24hr clock) but the old Backup package uses AM/PM but something
perhaps doesn't match between what is expected to be keyed in and
what the equations want/need afterwards. Also what is acknowledged
afterwards right at the top of the screen...

Or maybe I'm just too tired;~/

----best wishes, Robert

Offline Tib

  • *
  • 571
  • +0/-0
    • http://www.tibors.net
[Announce] Dar - Disk Archive (alpha build)
« Reply #40 on: September 20, 2006, 03:09:16 AM »
icpix

Have a look at the bug tracker ... you can follow whats been hapening.

http://bugs.contribs.org/show_bug.cgi?id=1880

Regards,

Tib

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #41 on: September 20, 2006, 03:24:36 AM »
jpl----

I left DAR set for 1.00AM and so off it went...

Using mc I see /mnt/smb with temporary folder and a domain folder.
At the moment the temporary path contains some.dar archives. There
is nothing at all in the domain folder other than an empty /set2 folder.
Certainly nothing yet appeared on the destination share of the LAN.
So, DAR *is* aggregating its backups of my server ON the server.
I don't know at what point or when it will transfer them to the given
destination drive ie over the LAN. That it is NOT doing this, as and
when system logistics allows, is disappointing because this has poor
implications for a site (mine) with up to 2TB dataset to backup.

Half way through the given 1hr backup window (if it works) and there
are now some five .dar archives in the temporary folder area but
still occupying space on the server's smallest drives. There is nothing
on the target/destination drive.

Having 'fixed' my second CPU (thanks to byte/Tib) I manually set off a
ClamAV run concurrently to the backup run. Now both CPUs are at peak
activity with the average CPU utilisation showing as 100% (htop). It's
getting warmer around here;~)

At the end of the allotted hour DAR *did* actually stop;~)
There are ten and a half .dar archives sitting in the domain path area
under the /set2 folder. There is *nothing* on the target drive across
the LAN but I don't know if that is because they are *now* in transit
or whether DAR might not have been able to write on the target(?).

<see if the shared folder is really mounted and if the server can write in it ?>
I don't know how to properly check what you have asked.

It's late (2.30am) - been a long day - much activity - g'night;~)

----best wishes, Robert

icpix

[Announce] Dar - Disk Archive (alpha build)
« Reply #42 on: September 20, 2006, 03:29:55 AM »
Tib----
Yes, I dive in and look sometimes.
It's very technical in there, mostly too difficult for me;~/
I stay away from it if I can;~)
The glitch (for jpl----) about the witching hour is different to
what was discussed about the merits of AM/PM and that of
matching the flavour of the existing Backup & Restore package.
Always match - Never advance - Never fix... is my view.
Should Always improve - Assessed each time.
(Many thanks again for the help in the dual CPU thread)
----best wishes, Robert (g'night)

Offline jpl

  • *
  • 112
  • +0/-0
[Announce] Dar - Disk Archive (alpha build)
« Reply #43 on: September 20, 2006, 03:54:11 PM »
Quote from: "icpix"
jpl----

Using mc I see /mnt/smb with temporary folder and a domain folder.
At the moment the temporary path contains some.dar archives. There
is nothing at all in the domain folder other than an empty /set2 folder.
Certainly nothing yet appeared on the destination share of the LAN.
So, DAR *is* aggregating its backups of my server ON the server.


There is only one reason that can make dar write on your server : the share  is not mounted correctly.

Normally, the script could detcet it, but it seems the rpm lacks creation of one directory.

Please :

- on your server, delete all directories under /mnt/smb (be sure the share is not mounted, ie after a reboot)
- create the directory /mnt/smb/smbmount-failure

This will prevent dar to do backups on the server itself.

Quote

I don't know at what point or when it will transfer them to the given
destination drive ie over the LAN.


Never, because the temporary directory should be on the destination system, not on the server (and would be in case of correct mount of the share).

Quote
At the end of the allotted hour DAR *did* actually stop;~)
There are ten and a half .dar archives sitting in the domain path area
under the /set2 folder.

Good in itself. If the share had been mounted, all where ok and the backup session would be correct.

Quote

There is *nothing* on the target drive across
the LAN but I don't know if that is because they are *now* in transit
or whether DAR might not have been able to write on the target(?).

There are not in transit. The explenation is the same : they would have been on the target if the shared had been correctly mounted.

Offline dmay

  • *
  • 450
  • +0/-0
    • http://myezserver.com
[Announce] Dar - Disk Archive (alpha build)
« Reply #44 on: September 20, 2006, 04:51:58 PM »
The original post stated all feedback should go to the Bugzilla ticket. All your problems encountered have been addressed and answered in the ticket. FYI this is an alpha build that will change substantially from day-to-day as new feedback is received, problems identified and fixed and new ideas implemented. If you are keen to play along, the playpen is Bugzilla  :wink:  Not here.

Darrell