Excessive limitation checks and logging when Custom Peak time below limit of active insulin curve

[mountrcg]

Describe the bug
When running on an insulin curve = rapid-acting and enabling a custom peak time of 48 min, which is under the defined minimum of 50mg/dl excessive logging will start at every loop displaying :
JavaScript log: Setting minimum Insulin Peak Time of 50m for,rapid-acting,insulin
150 times every second for up to 60s resulting in 75% failed loops.
This happened when I switched back from Lyumjev and missed disabling the custom peak time.

Expected behavior
change once and display once per loop cycle

Setup Information (please complete the following information):

• Pump type Omnipod
• CGM type G6
• Rig type iPhione
• oref0 v0.7.1-dev on FreeAPS-X

Additional context
the logging happens in the denomination with OpenAPS, so it is a result of determine-basal in conjunction with the other js-files meal iob etc, where I found above mentioned string.

[scottleibrand]

If you’re seeing this in FreeAPS-X, you should probably open an issue there, and once the problem is identified and fixed, see if any changes need to be upstreamed here to oref0.

Alternatively you could try to reproduce the problem in oref0 itself, but based on your hardware I think you’d have compatibility issues trying that.

[mountrcg]

FreeAPS-X just uses the standard OpenAPS, I think the problem should arise on pi-Rig as well. Yes I cannot test, can you? Reproduction should be easy

curve: "rapid-acting" ,
useCustomPeakTime: true,
insulinPeakTime: 48,

[ChanceHarrison]

This issue is partially reproducible on my Pi rig. I see the log spam in /var/log/openaps/shared-node.log but you wouldn't know it from looking at /var/log/openaps/pump-loop.log. It doesn't cause any loop failures on the Pi. I am working on a fix.

[ChanceHarrison]

This issue can be closed; it would have been closed automatically but the PR was "closed" instead of "merged" despite the changes being committed to dev.

[scottleibrand]

Per #1437 (comment) I had to revert this fix via 10f9e27

The new code was causing future IOB predictions to be incorrect (highly negative).
That causes the predBG lines to swoop up and to the right.
That results in (slightly) too much insulin being dosed in some situations.

[mountrcg]

@ChanceHarrison or @scottleibrand is there another solution on the horizon?

[scottleibrand]

I am not working on the issue. Would you like to?

[mountrcg]

No unfortunatly those interaction with the other files are above my laymans skill level.