Hi Jürgen,
Thanks for your reply.
It appears that this isn't a general problem so must be something with my
setup. I don't think it's a database issue or disk IO issue - typical iostat
output is
avg-cpu: %user %nice %sys %iowait %idle
23.58 0.02 0.32 1.35 74.72
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
sda 11.40 163.20 23.90 1632 239
sdb 0.00 0.00 0.00 0 0
hda 0.00 0.00 0.00 0 0
I'm not really sure what else to try.
I'm running 2.2.8, using PostgreSQL 8.1 as the database on SuSE 10.1
Any other ideas?
Thanks,
Rob
On Tue, Apr 08, 2008 at 12:10:20PM +0200, J?rgen Kuri wrote:
> Hello Rob,
>
> I encounterd the nearly same phenomenon. I tried to made a restore of ~1,4
> million jobfiles. I had not the irresponsiveness of bconsole after the
> marking of that directory - this was quite fast - as you but after typing
> "done" in the file selection context. Then I encountered the very same
> phenomenon as you: the bconsole program consumed one enitre CPU-core for
> hours and there was no database activity. I am testing bacula at the moment
> again and have this issue with bacula release 2.2.8 and three years ago - in
> an another test - with release 1.36.
>
> I consulted this forum too but I got no solution. I ran the director daemon
> in debug mode but I could not interprete the messages. So, one way to
> encounter the problem is to debug the director and take a deeper look into
> the source code of the director, but I am not familiar with C/C++ - that's my
> problem.
>
> I don't know how we can attract the attention of the developers. Maybe your
> ticket will induce it.
>
> See also the thread to my consult to this forum:
> http://marc.info/?l=bacula-users&m=120602534417470&w=2
>
>
> J?rgen
>
> -----Urspr?ngliche Nachricht-----
> Von: bacula-users-bounces AT lists.sourceforge DOT net
> [mailto:bacula-users-bounces AT lists.sourceforge DOT net] Im Auftrag von Rob
> Horton
> Gesendet: Dienstag, 8. April 2008 10:04
> An: bacula-users AT lists.sourceforge DOT net
> Betreff: [Bacula-users] Console very slow when restoring large directories
>
> Hi,
>
> I've noticed that the console program becomes unresponsive for a long
> time when marking a largeish directory for restore. For example, for
> a directory with 800,000 files it took four hours between typing
> "mark directory" and getting a prompt to actually start the job (the
> actual restore only took about 1 hour). This is obviously going to
> be a major problem when restoring very large directories as it would
> take days before it even starts reading the tape. The time seems to
> scale roughly linearly with the number of files marked.
>
> Has anybody else noticed this? Is there a way round this I haven't
> spotted?
>
> I don't think the problem is the database access speed, the director
> appears to be CPU-bound.
>
> Thanks,
> Rob
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
> Register now and save $200. Hurry, offer ends at 11:59 p.m.,
> Monday, April 7! Use priority code J8TLD2.
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
> _______________________________________________
> Bacula-users mailing list
> Bacula-users AT lists.sourceforge DOT net
> https://lists.sourceforge.net/lists/listinfo/bacula-users
>
-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Register now and save $200. Hurry, offer ends at 11:59 p.m.,
Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone _______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|