#give-feedback
Title
# give-feedback
b
busy-horse-73824
11/20/2023, 11:47 PMWe've just found that one of our SDK connections' project filter has disappeared? Caused quite a major bug for all users visiting our site. Has anyone else seen anything like that?
ID sdk_19g61pldef9nwj
(I have since re-set the project filter)
If this was a bug on GB's side, we could do with knowing exactly when it happened so we can identify the downtime impact
👀
f
future-teacher-7046
11/21/2023, 12:00 AM
Yes, looking at it now. We should have migrated all of the old projects settings over to the new format
b
busy-horse-73824
11/21/2023, 12:00 AM👌
f
future-teacher-7046
11/21/2023, 12:00 AM
not sure why that didn't happen
Ok, found the bug and have a fix being deployed in a few minutes. https://github.com/growthbook/growthbook/pull/1869
All of the old project settings should be restored now. Sorry about that. Our local testing of that PR didn't account for some production-specific MongoDB settings. We'll make changes to our CI/CD pipeline to test for this going forward to avoid any other issues like this in the future.
👌 2
b
busy-horse-73824
11/22/2023, 3:38 PMThank you for sorting this 🙂
One quick concern - the Growthbook JS SDK has client-side caching in place (with localstorage). Do you know how long their TTL is, or how the cache can be busted for all uses of the SDK key?
It seems that some of our users are still having the problem as of a few mins ago 😬
I guess worst-case I could just make a new SDK connection, but I wonder if this is affecting any other Growthbook tenants too
f
future-teacher-7046
11/22/2023, 4:43 PM
We use local storage cache to initialize the SDK, but if the cached data is more than 60 seconds old, we then do a network request to refresh it.
One thing you could do is change the cache key to wipe out any old data.
Copy code
import {configureCache} from "@growthbook/growthbook";
configureCache({
cacheKey: "gbFeaturesCache",
});gbFeaturesCacheJust make sure to call
configureCacheb
busy-horse-73824
11/22/2023, 5:22 PMahh we were triggering a redirect after ~50ms, so seems likely it was just never able to update the cache
We're removing our "no code redirect" setup for now, it's a bit too risky really
Thanks 🙂
3 Views