Add fan_dev_shm_enable signal to Odin

[jacob720]

Needed for DiamondLightSource/mx-bluesky#1076
We are currently migrating to the ophyd-async Eiger, and trying to use this in Hyperion. We need to be able to turn on dev/shm streaming for GPU analysis during xray centring, and turn it off to clear the memory, this PR adds the relevant PV to the Odin device.

[jacob720]

Hi, @GDYendell is this signal in the IOC?

[GDYendell]

Yep that looks right. For I03 it is BL03I-EA-EIGER-01:OD:FAN:DevShmCache

[jacob720]

Great, thanks

[DominicOram]

The B21 SAXS detector does not have this PV(though curiously the WAXS one does):

(.venv) [ffv81422@ws455 dodal]$ caget BL21B-EA-EIGER-01:OD:FAN:DevShmCache
Channel connect timed out: 'BL21B-EA-EIGER-01:OD:FAN:DevShmCache' not found.
(.venv) [ffv81422@ws455 dodal]$ caget BL21B-EA-EIGER-02:OD:FAN:DevShmCache
BL21B-EA-EIGER-02:OD:FAN:DevShmCache No

@GDYendell, I know this PV was always meant to be temporary but I don't think we're in a position where we can remove it soon? Can we add it to the B21 SAXS detector instead?

FYI @RJCD-Diamond

[RJCD-Diamond]

The B21 SAXS detector does not have this PV(though curiously the WAXS one does):

(.venv) [ffv81422@ws455 dodal]$ caget BL21B-EA-EIGER-01:OD:FAN:DevShmCache
Channel connect timed out: 'BL21B-EA-EIGER-01:OD:FAN:DevShmCache' not found.
(.venv) [ffv81422@ws455 dodal]$ caget BL21B-EA-EIGER-02:OD:FAN:DevShmCache
BL21B-EA-EIGER-02:OD:FAN:DevShmCache No

@GDYendell, I know this PV was always meant to be temporary but I don't think we're in a position where we can remove it soon? Can we add it to the B21 SAXS detector instead?

FYI @RJCD-Diamond

This is related to: #1117 annoyingly the detector on b21 that doesn't have the nexpi PV and the DevShmCache is the newer detector

[GDYendell]

The PV isn't temporary. What might change is that we don't rely on toggling it off and on to clear the memory when the consuming process deletes the images as soon as they have been processed. Although it would probably still needed in some cases like scan failure where there might be some images left behind.

[RJCD-Diamond]

@jsouter Are you able to add DevShmCache and nexpi to the b21 eiger's?

[jsouter]

@jsouter Are you able to add DevShmCache and nexpi to the b21 eiger's?

The PVs are dynamically created by fastcs-eiger directly from the API as reported by the eiger. One of the Eigers on B21 (I think SAXS?) does not provide nexpi, I will double check whether that is just because it's missing from the keys attribute of the detector or if the control is actually absent from the API. I'll have to look into DevShmCache

EDIT: nexpi is missing from the SAXS detector (Eiger 1 Si 4M)
EDIT2: I believed both the SAXS and WAXS were running of the same version of odin but that's not the case, I can try bumping ADOdin to 0-21-beta4 for the SAXS detector on tuesday, that should add in DevShmCache

[coretl]

@DominicOram @jacob720 I thought we were going to delete ADOdin from ophyd-async and switch testing FastCS Odin instead? That would mean this change would be deleted shortly...

[DominicOram]

@DominicOram @jacob720 I thought we were going to delete ADOdin from ophyd-async and switch testing FastCS Odin instead? That would mean this change would be deleted shortly...

We're actively testing the FastCS Eiger + ADOdin in the meantime with this patched on to make sure Hyperion works with FastCS Eiger / StandardDetector generally. Once #1158 is merged then happy to drop this