one machine out of 10 will not read .vdbs in a volumeGrid

[smbell]

so, like the title says. We have one machine that is configured identically to 9 others that just won't read certain .vdbs. It says:

[2019/Jan/4|14:23:05] V-Ray: [PhoenixCachePlugin] Frame data 141000 successfully loaded from \\akashi\proj\apollo11\cache\fumefx\faster_main_bo oster_4b5\faster_main_booster_4b5_####.vdb.

[2019/Jan/4|14:23:05] V-Ray: [PhoenixSimPlugin] grid 185x1248x185, cache index -2147483648.

[2019/Jan/4|14:23:06] V-Ray: [PhoenixCachePlugin] Frame data 1240000 can't be loaded from \\akashi\proj\apollo11\akashi\warehouse\apollo11\c ache\fumefx\left_smoke_4\left_smoke_4_pp_#.vdb.

[2019/Jan/4|14:23:06] V-Ray: [PhoenixSimPlugin] No simulation data for frame time 2340!

[2019/Jan/4|14:23:06] V-Ray: [PhoenixCachePlugin] Frame data 1240000 can't be loaded from \\akashi\proj\apollo11\cache\fumefx\right_smoke_3_ 1.0\right_smoke_3_1.0_#.vdb.

[2019/Jan/4|14:23:06] V-Ray: [PhoenixSimPlugin] No simulation data for frame time 2340!

so it reads one from a network share, then it doesn't read another a different one. If I open maya on the machine, the volumegrid vdb selection input is blank. I can pick the .vdb and it loads, then if i save the scene and render it doesn't load. These shares are on the same centos 7 box using samba, and like i said, all the other machines are perfectly fine.

Our VRay installation is loading from the network, so all the machines are using the same version, the hotfix 1 next. The maya versions are all 2018.4.

Is there anything else anyone can think of that would cause this. It's driving us crazy.

[georgi.zhekov]

From what I can see the caches that are not read are using only one # symbol at the end while the first cache is using four so if you're not using Cache index mode in the Input rollout it won't be able to find those caches.
Does it work if you change the paths to use #### instead of #?

[smbell]

unfortunately, no it still doesn't work with ####. It's happening on three different scenes. I'm using linear for the playback mode because we need the timeline origin to be slid around. It's just this one machine that has the problem, it's bizarre.

[georgi.zhekov]

Does the machine has access to the network? Is reading textures from the same location working?