You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is not a security vulnerability or a crashing bug
This is not a question about how to use Istio
Bug Description
Running Istio in multi-primary setup.
When using localityLbSetting enabled with outlierDetection traffic is being routed to local cluster only, but some requests are getting timeout errors.
When either a) outlierDetection is removed and localityLbSetting is enabled or b) outlierDetection is present and localityLbSetting is disabled, traffic is being balanced between two clusters
Screenshots of application connection timeout error logs and proxy sidecars metrics shows the correlation when outlierDetection is removed - errors disappear, but traffic is being balanced between two clusters.
Interestingly enough, localityLbSetting enablement adds to EnvoyProxy clusters config:
commonLbConfig:
localityWeightedLbConfig: {}
While adding outlierDetection, under the hood healthyPanicThreshold is being added to EnvoyProxy config:
It seems that the issue is application specific. When application itself has timeouts and handles retries, EnvoyProxy on top does its own magic. Turning off application timeouts and retries and setting those with VirtualService solved our issue.
Is this the right place to submit this?
Bug Description
Running Istio in multi-primary setup.
When using
localityLbSetting
enabled withoutlierDetection
traffic is being routed to local cluster only, but some requests are getting timeout errors.When either a)
outlierDetection
is removed andlocalityLbSetting
is enabled or b)outlierDetection
is present andlocalityLbSetting
is disabled, traffic is being balanced between two clustersScreenshots of application connection timeout error logs and proxy sidecars metrics shows the correlation when
outlierDetection
is removed - errors disappear, but traffic is being balanced between two clusters.Interestingly enough,
localityLbSetting
enablement adds to EnvoyProxy clusters config:While adding
outlierDetection
, under the hoodhealthyPanicThreshold
is being added to EnvoyProxy config:Additionally all endpoints have
priority
assigned to them.I'm interested what magic does
outlierDetection
andlocalityLbSetting
combo adds that outgoing requests starting to time out?Version
Additional Information
No response
The text was updated successfully, but these errors were encountered: