Jeff Schroeder wrote:
> On Wed, May 6, 2009 at 11:49 AM, Michael DeHaan <[email protected]> wrote:
>   
>> The problem Smooge reported with cobblerd yesterday evening deals with
>> cobblerd doing some undesirable things when rename commands are made in
>> the CLI.
>> AFAIK, this is a 1.6.3 bug but may be present in earlier versions.
>>
>> This will be included in 1.6.4, out Thursday or Friday, but is included
>> here for completeness.
>>
>> Users should, without question, apply this patch, if they use the CLI
>> and will ever use, intentionally or otherwise, the "rename" command:
>>
>> *http://tinyurl.com/crkgkx
>> *
>> There may be some further fixing to the logic behind this for 1.6.4,
>> though we've confirmed this takes care of the problem.
>>     
>
> How did you find this? RTFC? Nice work.
>
>   

Same way I broke it by WTFC, I guess.

It seemed to be showing recursive properties -- messing with a 
subobject. There's very little that could have been doing that, 
especially with the deltas in the timestamps.

The real catch is the code still says "delete=False", so the fix is 
valid, but not fundamentally the perfect way to fix it -- still 
exploring. But there's no need for a delete following a rename to be 
recursive, because it's just one object being deleted, and a delete via 
cobbler notification happens one-per-object anyway (and thus does not 
need to be recursive). Not that I really expect folks to need to 
understand that, I don't.

So it's a fix, but not the ideal fix. The ideal fix is understanding why 
the recursive delete happened when we didn't ask it to.

--Michael
_______________________________________________
cobbler mailing list
[email protected]
https://fedorahosted.org/mailman/listinfo/cobbler

Reply via email to