Erroneous Last full dump overwritten message

2004-07-08 Thread KEVIN ZEMBOWER
I go this message last night:
NOTES:
  planner: Last full dump of www:sda11 on tape  overwritten in 1 run.

And yet, I have plenty of level 0 backups:
[EMAIL PROTECTED]:~$ amadmin Outside find www sda11
Scanning /dumps/amanda...

date   host disk  lv tape or file file status
2004-06-09 www  sda11  2 Outside-21 10 OK
2004-06-11 www  sda11  3 Outside-01 12 OK
2004-06-12 www  sda11  0 Outside-02 15 OK
2004-06-15 www  sda11  1 Outside-03 12 OK
2004-06-16 www  sda11  2 Outside-04 13 OK
2004-06-17 www  sda11  3 Outside-05 11 OK
2004-06-19 www  sda11  3 Outside-07 14 OK
2004-06-22 www  sda11  0 Outside-08 15 OK
2004-06-23 www  sda11  1 Outside-09 13 OK
2004-06-24 www  sda11  2 Outside-10 13 OK
2004-06-25 www  sda11  3 Outside-11  9 OK
2004-06-26 www  sda11  0 Outside-12 16 OK
2004-06-29 www  sda11  1 Outside-13 12 OK
2004-06-30 www  sda11  2 Outside-14 11 OK
2004-07-01 www  sda11  3 Outside-15 10 OK
2004-07-02 www  sda11  3 Outside-16 12 OK
2004-07-03 www  sda11  3 Outside-17 11 OK
2004-07-07 www  sda11  0 --- 0 FAILED (driver) [dump to tape failed]
2004-07-07 www  sda11  0 --- 0 FAILED (dumper) [data write: Connection 
reset by peer]
2004-07-07 www  sda11  0 Outside-19 14 [out of tape]
2004-07-08 www  sda11  0 Outside-20 16 OK
[EMAIL PROTECTED]:~$ 

This indicates that one was just made last night, but three others existed within the 
tapecycle.

My amanda.conf setting are:
dumpcycle 8 # the number of days in the normal dump cycle
runspercycle 8  # the number of amdump runs in dumpcycle days
tapecycle 15 tapes  # the number of tapes in rotation

Is this message indicating that there's no level 0 backup within the last dumpcycle 
tapes? 

This isn't a big concern; I've been just ignoring it for a couple of years now. But 
today, I'm curious and have the time to write about it.

Also, is there something missing after the word 'tape' in the message, planner: Last 
full dump of www:sda11 on tape  overwritten in 1 run.? On my other operating amanda 
system, this is usually filled in by the tape name.

Thanks for your thoughts and suggestions.

-Kevin

-
E. Kevin Zembower
Unix Administrator
Johns Hopkins University/Center for Communications Programs
111 Market Place, Suite 310
Baltimore, MD  21202
410-659-6139




Re: Erroneous Last full dump overwritten message

2004-07-08 Thread Alexander Jolk
KEVIN ZEMBOWER wrote:
 Also, is there something missing after the word 'tape' in the message, planner: 
 Last full dump of www:sda11 on tape  overwritten in 1 run.? On my other operating 
 amanda system, this is usually filled in by the tape name.

I've seen the same thing here, and I believe it is caused by a dump
that's been forgotten on holding disk and won't be flushed.  Can happen
when you do several amdumps on the same day, it seems.

Sound plausible to someone?

Alex

-- 
Alexander Jolk / BUF Compagnie
tel +33-1 42 68 18 28 /  fax +33-1 42 68 18 29


Re: Erroneous Last full dump overwritten message

2004-07-08 Thread Paul Bijnens
Alexander Jolk wrote:
KEVIN ZEMBOWER wrote:
Also, is there something missing after the word 'tape' in the
message, planner: Last full dump of www:sda11 on tape  overwritten
in 1 run.? On my other operating amanda system, this is usually
filled in by the tape name.

I've seen the same thing here, and I believe it is caused by a dump 
that's been forgotten on holding disk and won't be flushed.  Can
happen when you do several amdumps on the same day, it seems.

Sound plausible to someone?
That's correct.
It's a little bug in the code.
Creating a little patch is on my todo list (when I find the time).
The bug is in this little subroutine in planner.c.  Something that
shouldnt happen does happen:  when the last full dump is
on holdingdisk, then the routine gets called with an empty label,
and it returns 1 to trigger warning message.
 925 static int when_overwrite(label)
 926 char *label;
 927 {
 928 tape_t *tp;
 929
 930 if((tp = lookup_tapelabel(label)) == NULL)
 931 return 1;   /* shouldn't happen, but trigger warning 
message */
 932 else if(!reusable_tape(tp))
 933 return 1024;
 934 else if(lookup_nb_tape()  conf_tapecycle)
 935 return (lookup_nb_tape() - tp-position) / conf_runtapes;
 936 else
 937 return (conf_tapecycle - tp-position) / conf_runtapes;
 938 }
 939

You see where the empty tape name comes from and the overwritten in 1
run too.
I didn't yet decide how to resolve it.  Is an empty label really
an indication of a dump on holdingdisk in all cases?  In that case
just return 1024 or so.
Or should we change infofile to use some string like **HOLDINGDISK**
as a fake tapelabel?
--
Paul Bijnens, XplanationTel  +32 16 397.511
Technologielaan 21 bus 2, B-3001 Leuven, BELGIUMFax  +32 16 397.512
http://www.xplanation.com/  email:  [EMAIL PROTECTED]
***
* I think I've got the hang of it now:  exit, ^D, ^C, ^\, ^Z, ^Q, F6, *
* quit,  ZZ, :q, :q!,  M-Z, ^X^C,  logoff, logout, close, bye,  /bye, *
* stop, end, F3, ~., ^]c, +++ ATH, disconnect, halt,  abort,  hangup, *
* PF4, F20, ^X^X, :D::D, KJOB, F14-f-e, F8-e,  kill -1 $$,  shutdown, *
* kill -9 1,  Alt-F4,  Ctrl-Alt-Del,  AltGr-NumLock,  Stop-A,  ...*
* ...  Are you sure?  ...   YES   ...   Phew ...   I'm out  *
***