All of the catalog
Timed challenge
On-call shift
You have just taken the pager. go-api has been down since the overnight maintenance, and the people who ran it have gone home. More than one change went out, and nobody wrote down which. Get go-api serving, and keep it serving: the shift ends when every change that broke it is fixed.
WorkloadPar 12m−5 points per hint
Definition on GitHubHow it is scored
- Par time is 12m. Finish inside it and time costs nothing. Going over costs up to 20 points, the further over the more it costs.
- Each hint costs 5 points out of 100. Taking one is often still the right call.
- Grading is a real check against the live cluster, so a fix that only looks right scores nothing.
What gets set up
Related challenges
- Restore the broken deployA release went out an hour ago and never finished. Users are still being served by the old version, the rollout has not advanced, and the deploy pipeline is red. Get the rollout healthy before the clock runs out.
- Find the memory leakThe ticket says "memory leak in go-api". What is known: the workload came up clean, served fine for a minute, and then began restarting over and over once traffic arrived. It is still restarting now. Stop it, and leave the workload able to serve the traffic it is getting.
- Make the SLO greengo-api's availability SLO is breaching: the error rate through the ingress is far above threshold and the burn rate is climbing. Every pod is Running and Ready, and nothing was deployed. Find the root cause and confirm the SLO recovers.