feat: Add grace period to uptime check #167
Conversation
| function groupUptimeChecksByPeriod(uptimeChecks: { isOnline: boolean; checkDate: string }[] = []) { | ||
| const groupedSnapshots: { checkDate: Date; checks: boolean[] }[] = []; | ||
|
|
||
| uptimeChecks.toSorted((a, b) => new Date(a.checkDate).getTime() - new Date(b.checkDate).getTime()); |
There was a problem hiding this comment.
Didn't know about this method :) after looking into the doc and testing it seems like it's not mutating the original array. So seems like this usage might not be intended
There was a problem hiding this comment.
Good catch, I changed the sort for toSorted to make use of the new immutable array methods which I prefer, but forgot to add the new variable. 😄 There's some other cool ones like toSpliced and toReversed
| const uptime1d = provider.total1d > 0 ? provider.online1d / provider.total1d : 0; | ||
| const uptime7d = provider.total7d > 0 ? provider.online7d / provider.total7d : 0; | ||
| const uptime30d = provider.total30d > 0 ? provider.online30d / provider.total30d : 0; | ||
| const uptime1d = Math.max(0, 1 - offline_seconds_1d / (24 * 60 * 60)); |
There was a problem hiding this comment.
i don't mind this numbers. just dropping it here https://www.npmjs.com/package/time-constants - used to use it. a lib is quite simple but improves readability quite a bit
There was a problem hiding this comment.
Good idea, I found out that date-fns has similar constants so I used those instead. Way more readable 👍
| })); | ||
| } | ||
|
|
||
| const uptimePeriods = groupUptimeChecksByPeriod(provider?.uptime || []); |
There was a problem hiding this comment.
I think this method could be extracted as a hook with useMemo since it's called on every render and does a decent amount of calculations.
Redm4x
force-pushed
the
refactor/improve-provider-uptime-tracking
branch
from
ac7c09d
to
eafb3ec
Compare
Redm4x
force-pushed
the
refactor/improve-provider-uptime-tracking
branch
from
eafb3ec
to
2758a90
Compare
Improvements to the provider uptime tracking system #152
Added a grace period before treating a provider as completely offline. This will reflect in the following places:
/internal/gpu-pricesendpoint used by the GPU Pricing page/internal/gpu&/internal/provider-versionsendpoints used for internal trackingThe default grace period is 3h, but this can be changed with the
ProviderUptimeGracePeriodMinutesenv var in the api.When a provider fails an uptime check, start retrying every minutes and slowly increase the interval based on the following table: