On Wednesday 15 April 2009, Gene Heskett wrote:
>On Wednesday 15 April 2009, Gene Heskett wrote:
>>On Tuesday 14 April 2009, Jean-Louis Martineau wrote:
>>>I committed a fix, thank for finding the bug.
>>>
>>>CHECK-DEVICE is correct, properties are sent send to the 'support'
>>>application command because they are not needed, the script just log the
>>>default value.
>>>
>>>Jean-Louis
>>>
>>>Dustin J. Mitchell wrote:
>>>> OK, after some logfile exchange off-list, I think we're onto something
>>>> here.  The planner examines all DLEs to decide which incremental to
>>>> promote, and selects coyote:/root.  It finishes its loop, and then
>>>> goes to attempt to promote that dump:
>>>>
>>>> 2792     if(dp_promote) {
>>>> 2793         one_est_t *level0_est = est_for_level(dp, 0);
>>>> 2794         dp = dp_promote;
>>>>
>>>> Sadly, dp is NULL on line 2793.  The fix is to use dp_promote instead
>>>> of dp.  Jean-Louis, do you agree?  If so, please commit (I'm running
>>>> off for the evening).
>>>>
>>>> Also, Jean-Louis, I'm leaving the chasing-down of the CHECK-DEVICE
>>>> question to you.
>>>>
>>>> Gene: thanks for working so hard on this, and my apologies if I seemed
>>>> short with you.  In general, when you see something go wrong with
>>>> Amanda, it's probably best to freeze everything and get to the bottom
>>>> of it immediately.
>>>>
>>>> Dustin
>>
>>Not so sure about this backup, started at 01:45, it is still running at
>> 5:01 am, apparently getting stuck waiting for the last 4kb of a dle.  But
>> it will eventually get it and go on.  My algorithm setting is largest
>> first. amstatus's extra output lines aren't really a help when scrolling
>> back to see where its actually at.
>>It gets here fairly quickly:
>>
>>coyote:/root                  2   1330791k dumping  1322112k ( 99.35%)
>>(4:57:50)
>>And it still reads the same at 5:20am.  And 26 dle's yet to go out of
>> thirty.
>>
>>Now there are no tar's running, and gzip is idling at 0 cpu, and will
>>apparently be stuck there till something times out. From amanda.conf:
>>[root@coyote /]# grep dtimeout /usr/local/etc/amanda/Daily/amanda.conf
>>dtimeout 2800           # number of idle seconds before a dump is aborted.
>>
>>But amstatus isn't reporting any aborts.  Of those 3 done 'finished'
>>[root@coyote log]# amstatus Daily|grep coyote|grep finished
>>coyote:/home                  0   2217790k finished (4:36:48)
>>coyote:/usr/pix               0   2158180k finished (4:58:56)
>>coyote:/usr/src               0   3139051k finished (3:49:24)
>>
>>So something is amiss. However it is progressing, albeit slowly.  I'm going
>>back to bed.  Need more zz's.
>
>PS: At 9:40 am, 14 done, one in progress, 15 to go.

Another PS: At 13:51 it is finally done, so instead of an hour 15 minutes, it 
took 12 hours & change.  This was amanda-2.6.2alpha-20090414. 415 is on hand 
and will be tested tonight.  I am also going to cut the dtimeout to half.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
<Dr^Nick> SGI_Multitexture is bad voodoo now
<Dr^Nick> ARB is good voodoo
<witten> no, voodoo rush is bad voodoo :)