Re: [HTCondor-users] purpose of SHADOW_WORKLIFE option

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

On 3/27/2014 1:31 PM, Brian Bockelman wrote:
> There's no point to have shadow worklife shorter than the claim, right?

There is no point unless you are worried about the shadow leaking some 
resource (memory? descriptors?) between jobs, or if you want to do 
something exotic like set things up so you have a separate shadow log 
file per job. FWIW, we are not aware of any such resource leak bugs in 
the shadow ...

> Are there any strange effects if shadow worklife is significantly longer than claim worklife?

None I am aware of... but be aware that as currently implemented, 
shadows will only get reused for the same claim.  So shadow worklife is 
effectively bounded by claim worklife.

Todd

--
Todd Tannenbaum <tannenba@xxxxxxxxxxx> University of Wisconsin-Madison
Center for High Throughput Computing   Department of Computer Sciences