@jdotjdot installed Pixie and is seeing the following issues:
px CLI says the cluster is unhealthy much of the time
when running scripts, often gets a “stream read error”
@nserrino took a quick look at the logs: "It looks like the PEMs are taking more than 5 seconds to send their results to the Kelvin, which shouldn't be happening. Looks like kelvin is correctly sending health check results back, but query broker is interpreting them as invalid for some reason."
App information (please complete the following information):
self-hosted on AWS using kops. v1.18.8
we use spotinst for control plane / scheduling
zasgar
commented
3 years ago
@jdotjdot installed Pixie and is seeing the following issues:
@nserrino took a quick look at the logs: "It looks like the PEMs are taking more than 5 seconds to send their results to the Kelvin, which shouldn't be happening. Looks like kelvin is correctly sending health check results back, but query broker is interpreting them as invalid for some reason."
App information (please complete the following information): self-hosted on AWS using kops. v1.18.8 we use spotinst for control plane / scheduling