To measure the current problems with the server, I...
# github-workflows-ktp
Piotr Krzemiński
06/01/2025, 3:55 PMTo measure the current problems with the server, I thought about giving Grafana Synthetic Monitoring a shot. I see that it requires a Grafana Cloud account, but there's a free tier: https://grafana.com/products/cloud/. @LeoColman what do you think?
Piotr Krzemiński
06/01/2025, 4:00 PMI thought about a synthetic test that at least once a minute calls several endpoints, for several kinds or artifacts
l
LeoColman
06/01/2025, 4:59 PMI don't have an opinion on this. It's fine for me either way, knowing that we are going to depend a bit on a free tier that might not be free in the future
p
Piotr Krzemiński
06/01/2025, 7:55 PMI thought about it for a while and realized that Grafana would let us most likely only send HTTP requests. It would be better than nothing, but ideally we'd test how actually users call the service, i. e. through Kotlin Scripting. Leo, would you be fine with hosting a new, simple worker service that calls a certain .main.kts script periodically and emits a metric if running the script was successful or not?
l
LeoColman
06/01/2025, 8:16 PMi. e. through Kotlin Scripting. Leo, would you be fine with hosting a new, simple worker service that calls a certain .main.kts script periodically and emits a metric if running the script was successful or not?Absolutely. Should be an easy service to host
🦜 1
v
Vampire
06/01/2025, 9:34 PMand emits a metric if running the script was successful or not?Remember, the script execution in this case is successful, as it automatically retries!
p
Piotr Krzemiński
06/02/2025, 4:11 AMPartially right - sometimes the retries aren't enough and the execution fails. So we should perhaps both have a Kotlin script that calls the server (to reflect the actual usage) and also make the HTTP calls directly
v
Vampire
06/02/2025, 6:50 AMHm, ok.
I've not seen it fail due to this so far.
p
Piotr Krzemiński
06/02/2025, 6:51 AMPiotr Krzemiński
06/02/2025, 6:52 AMI thought all the time you meant failures so frequent that they caused the script execution failures
v
Vampire
06/02/2025, 7:00 AMNo, I've always seen just one to three retries when manually looking and it was successful eventually and I also always clearly stated that. 🤷♂️
But even then it was a sign of some problem that could get worse with broader usage or over time.
👍 1