Hi! Hopefully this is the right place to ask, but ...
# ask-questionsm
microscopic-france-88490
01/29/2025, 10:19 PMHi! Hopefully this is the right place to ask, but I was hoping to get some help troubleshooting a recent experiment.
My team started an experiment in Growthbook with a 50% / 50% traffic split between control and variant; however, after starting it we realized all traffic was being routed to the variant. We did some internal investigation and everything looks correct on our side. We tried re-running it again a few days later and had the same issue.
Any ideas on whether there is something on the Growthbook side that could be impacting this?
b
brief-honey-45610
01/30/2025, 3:47 PMHi Annaliese, we'd need to see a screenshot of the Experiment's Overview and Results tabs in order to help troubleshoot. Could you post those here?
f
fresh-football-47124
02/02/2025, 3:38 AM
Can you share the code you're using? It could be that you're using a single ID for everyone, which would assign all to one of the variants
c
careful-memory-25657
02/07/2025, 7:56 PMHi all, I’m an engineer from Annaliese’s team that works on our experimentation. Here’s a screenshot of the experiment overview on the experiment that had all traffic only go to the variant:
I was watching the traffic in real time populate our database and every request was only going to the variant so we shut it down. This was the second time we launched this feature/experiment with this result
careful-memory-25657
02/07/2025, 7:58 PMThe
external_idsensible_<randomly generated UUID> Copy code
sensible_98be1076-a728-4470-890e-ca222e67b0fd
sensible_add5d48d-115b-4e75-b685-201877d08bfd
sensible_0c3deee5-a84e-4908-9fca-552e40c1742d
sensible_3b0c241c-54c6-4628-9f2f-9b5c94a814baf
fresh-football-47124
02/08/2025, 5:02 PM
can you share the SDK code (how you're calling the tracking callback, and setting attributes) and the code for the feature check?
c
careful-memory-25657
02/10/2025, 6:53 PMSure! Here are two screenshots of code. This is used as middleware in our Golang api which adds a config override field to a given request body if it’s applicable to an experiment. At first I thought something may have changed with the hashing algorithm but it seems like all of our past experiments on this same traffic used the hashing algorithm v2 in Growthbook so that doesn’t seem to be the reason for this behavior
m
microscopic-france-88490
02/20/2025, 9:43 PMHi @fresh-football-47124, any ideas on what may be causing this? We're hoping to use Growthbook for this experiment soon. Thank you!
19 Views