consul/test/integration/connect/envoy/case-prometheus/verify.bats

52 lines
1.5 KiB
Plaintext
Raw Normal View History

#!/usr/bin/env bats
load helpers
@test "s1 proxy admin is up on :19000" {
retry_default curl -f -s localhost:19000/stats -o /dev/null
}
@test "s2 proxy admin is up on :19001" {
retry_default curl -f -s localhost:19001/stats -o /dev/null
}
@test "s1 proxy listener should be up and have right cert" {
assert_proxy_presents_cert_uri localhost:21000 s1
}
@test "s2 proxy listener should be up and have right cert" {
assert_proxy_presents_cert_uri localhost:21001 s2
}
test: for envoy integration tests, wait until 's2' is healthy in consul before interrogating envoy (#6108) When the envoy healthy panic threshold was explicitly disabled as part of L7 traffic management it changed how envoy decided to load balance to endpoints in a cluster. This only matters when envoy is in "panic mode" aka "when you have a bunch of unhealthy endpoints". Panic mode sends traffic to unhealthy instances in certain circumstances. Note: Prior to explicitly disabling the healthy panic threshold, the default value is 50%. What was happening is that the test harness was bringing up consul the sidecars, and the service instances all at once and sometimes the proxies wouldn't have time to be checked by consul to be labeled as 'passing' in the catalog before a round of EDS happened. The xDS server in consul effectively queries /v1/health/connect/s2 and gets 1 result, but that one result has a 'critical' check so the xDS server sends back that endpoint labeled as UNHEALTHY. Envoy sees that 100% of the endpoints in the cluster are unhealthy and would enter panic mode and still send traffic to s2. This is why the test suites PRIOR to disabling the healthy panic threshold worked. They were _incorrectly_ passing. When the healthy panic threshol is disabled, envoy never enters panic mode in this situation and thus the cluster has zero healthy endpoints so load balancing goes nowhere and the tests fail. Why does this only affect the test suites for envoy 1.8.0? My guess is that https://github.com/envoyproxy/envoy/pull/4442 was merged into the 1.9.x series and somehow that plays a role. This PR modifies the bats scripts to explicitly wait until the upstream sidecar is healthy as measured by /v1/health/connect/s2?passing BEFORE trying to interrogate envoy which should make the tests less racy.
2019-07-10 20:58:25 +00:00
@test "s2 proxy should be healthy" {
assert_service_has_healthy_instances s2 1
}
@test "s1 upstream should have healthy endpoints for s2" {
assert_upstream_has_endpoints_in_status 127.0.0.1:19000 s2.default.primary HEALTHY 1
}
@test "s1 upstream should be able to connect to s2 with http/1.1" {
run retry_default curl --http1.1 -s -f -d hello localhost:5000
[ "$status" -eq 0 ]
[ "$output" = "hello" ]
}
@test "s1 proxy should be exposing metrics to prometheus" {
# Should have http metrics. This is just a sample one. Require the metric to
# be present not just found in a comment (anchor the regexp).
retry_default \
must_match_in_prometheus_response localhost:1234 \
'^envoy_http_downstream_rq_active'
# Should be labelling with local_cluster.
retry_default \
must_match_in_prometheus_response localhost:1234 \
'[\{,]local_cluster="s1"[,}] '
# Should be labelling with http listener prefix.
retry_default \
must_match_in_prometheus_response localhost:1234 \
'[\{,]envoy_http_conn_manager_prefix="public_listener_http"[,}]'
test: for envoy integration tests, wait until 's2' is healthy in consul before interrogating envoy (#6108) When the envoy healthy panic threshold was explicitly disabled as part of L7 traffic management it changed how envoy decided to load balance to endpoints in a cluster. This only matters when envoy is in "panic mode" aka "when you have a bunch of unhealthy endpoints". Panic mode sends traffic to unhealthy instances in certain circumstances. Note: Prior to explicitly disabling the healthy panic threshold, the default value is 50%. What was happening is that the test harness was bringing up consul the sidecars, and the service instances all at once and sometimes the proxies wouldn't have time to be checked by consul to be labeled as 'passing' in the catalog before a round of EDS happened. The xDS server in consul effectively queries /v1/health/connect/s2 and gets 1 result, but that one result has a 'critical' check so the xDS server sends back that endpoint labeled as UNHEALTHY. Envoy sees that 100% of the endpoints in the cluster are unhealthy and would enter panic mode and still send traffic to s2. This is why the test suites PRIOR to disabling the healthy panic threshold worked. They were _incorrectly_ passing. When the healthy panic threshol is disabled, envoy never enters panic mode in this situation and thus the cluster has zero healthy endpoints so load balancing goes nowhere and the tests fail. Why does this only affect the test suites for envoy 1.8.0? My guess is that https://github.com/envoyproxy/envoy/pull/4442 was merged into the 1.9.x series and somehow that plays a role. This PR modifies the bats scripts to explicitly wait until the upstream sidecar is healthy as measured by /v1/health/connect/s2?passing BEFORE trying to interrogate envoy which should make the tests less racy.
2019-07-10 20:58:25 +00:00
}