Open b1conrad opened 3 years ago
No, it's likely not ctx:query
; that's just what I had been working with when the incident occurred. It has happened since. Restarting the engine and waiting and then starting it again and the rulesets and channels are back.
Seems to be a UI failure. Next time it happens, I will check the browser console.
Noticed this: when visiting a tab that displays in this way (i.e. "no rulesets" or "no channels" etc.) the pico-engine console displays no messages at all. No error message. No entries that indicate that the pico-engine received a query from the UI.
How does the UI, running in the browser, get the information it needs? It must be from the engine. Yet no query is logged. How does it reach the engine? Does it use localhost:3000
or does it use an IP address? Perhaps this happens when localhost is assigned a different IP address?
Doing one of the queries manually in the browser, .../sky/cloud/ckj212eyb0007gn2rd0tq2u4h/io.picolabs.pico-engine-ui/box, it just spins. It's like the engine isn't running any more
The pico-engine seems to degrade (at least in the developer UI) when
ctx:query
has been used.Workaround is to avoid its use until this is diagnosed and fixed. It’s possible that
wrangler:skyQuery
could be used as a replacement (although its implementation is slated to usectx:query
when possible (currently it useshttp:get
)).I haven't been able to produce a small triggering bit of code, but do have a snapshot of the PICO_ENGINE_HOME folder after it has occurred. Starting the pico-engine from this point shows the defects in the UI. These are picos with no rulesets and no channels (even though they had some of both before the incident).