Farm Rendering multiple state sets with lightcache automatically?

[im.thatoneguy]

I've got 3,000+ preview frames to render every night with lighting and GI from 40 cameras. Is this the best approach for rendering a bunch of state sets? I wrote a pre-render script:

curcam = try((getActiveCamera()).name) catch()
vr=renderers.current

cachePath = @"\\server\cache\"+curcam

if not doesfileexist (cachePath+".vrlmap") then (
vr.adv_irradmap_mode = 0 ) else ( vr.adv_irradmap_mode = 2 )

if not doesfileexist (cachePath+".vrmap") then (
vr.lightcache_mode = 0 ) else ( vr.lightcache_mode = 2 )

vr.lightcache_loadfilename = cachePath+".vrlmap"
vr.lightcache_savefilename = cachePath+".vrlmap"
vr.adv_irradmap_autoSaveFileName =  cachePath+".vrmap"
vr.adv_irradmap_loadFileName =  cachePath+".vrmap"
vr.adv_irradmap_saveFileName  =  cachePath+".vrmap"

My intent is for the first farm machine(s) to only calculate the light cache and Irradiance map once and then subsequent frames reuse it for preview renders. I've set "use camera path" for both so it should be temporally stable. Is there anything wrong with this approach or am I reinventing a wheel that already exists and is more robust?

[^Lele^]

We do not suggest using the path modes for the LC, and rather have each frame calculate individually.
3k subdivs and a higher retrace amount provide for temporal stability in the vast majority of cases (raising either, or both, a bit more would help in more complex situations).
Rendertime over the whole sequence may be higher (that's depending on how the LC was set up), but setup complexity, and the risks of mishaps, are considerably lower, and each frame can be re-rendered at will without having to recalculate a whole sequence.

Cfr. here, at the very bottom.