VictoriaMetrics/lib/promscrape/config_test.go
Zhu Jiekun e97e966f82
lib/promrelabel: follow-up for 8958cecad6
In the previous commit 8958cecad6
the default ports (80/443) were removed for both the `scrapeURL` and
`instance` label values for those targets without a port in
`__address__`. Different values in the `instance` label generate new
time series.

This commit reverts the changes made to the `instance` label. Now,
for those targets:
- `scrapeURL` will remain unchanged.
- The `instance` label value will include the default port.

https://github.com/VictoriaMetrics/VictoriaMetrics/issues/6792
2024-08-27 13:04:26 +02:00

1355 lines
33 KiB
Go

package promscrape
import (
"fmt"
"reflect"
"strings"
"testing"
"time"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/promauth"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/promrelabel"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/promscrape/discovery/gce"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/promutils"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/proxy"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/stringsutil"
)
func TestMergeLabels(t *testing.T) {
f := func(swc *scrapeWorkConfig, target string, extraLabelsMap, metaLabelsMap map[string]string, resultExpected string) {
t.Helper()
extraLabels := promutils.NewLabelsFromMap(extraLabelsMap)
metaLabels := promutils.NewLabelsFromMap(metaLabelsMap)
labels := promutils.NewLabels(0)
mergeLabels(labels, swc, target, extraLabels, metaLabels)
result := labels.String()
if result != resultExpected {
t.Fatalf("unexpected result;\ngot\n%s\nwant\n%s", result, resultExpected)
}
}
f(&scrapeWorkConfig{}, "foo", nil, nil, `{__address__="foo",__metrics_path__="",__scheme__="",__scrape_interval__="",__scrape_timeout__="",job=""}`)
f(&scrapeWorkConfig{}, "foo", map[string]string{"foo": "bar"}, nil, `{__address__="foo",__metrics_path__="",__scheme__="",__scrape_interval__="",__scrape_timeout__="",foo="bar",job=""}`)
f(&scrapeWorkConfig{}, "foo", map[string]string{"job": "bar"}, nil, `{__address__="foo",__metrics_path__="",__scheme__="",__scrape_interval__="",__scrape_timeout__="",job="bar"}`)
f(&scrapeWorkConfig{
jobName: "xyz",
scheme: "https",
metricsPath: "/foo/bar",
scrapeIntervalString: "15s",
scrapeTimeoutString: "10s",
}, "foo", nil, nil, `{__address__="foo",__metrics_path__="/foo/bar",__scheme__="https",__scrape_interval__="15s",__scrape_timeout__="10s",job="xyz"}`)
f(&scrapeWorkConfig{
jobName: "xyz",
scheme: "https",
metricsPath: "/foo/bar",
}, "foo", map[string]string{
"job": "extra_job",
"foo": "extra_foo",
"a": "xyz",
}, map[string]string{
"__meta_x": "y",
}, `{__address__="foo",__meta_x="y",__metrics_path__="/foo/bar",__scheme__="https",__scrape_interval__="",__scrape_timeout__="",a="xyz",foo="extra_foo",job="extra_job"}`)
}
func TestScrapeConfigUnmarshalMarshal(t *testing.T) {
f := func(data string) {
t.Helper()
var cfg Config
data = strings.TrimSpace(data)
if err := cfg.unmarshal([]byte(data), true); err != nil {
t.Fatalf("parse error: %s\ndata:\n%s", err, data)
}
resultData := string(cfg.marshal())
result := strings.TrimSpace(resultData)
if result != data {
t.Fatalf("unexpected marshaled config:\ngot\n%s\nwant\n%s", result, data)
}
}
f(`
global:
scrape_interval: 10s
`)
f(`
scrape_config_files:
- foo
- bar
`)
f(`
scrape_configs:
- job_name: foo
scrape_timeout: 1.5s
static_configs:
- targets:
- foo
- bar
labels:
foo: bar
`)
f(`
scrape_configs:
- job_name: foo
honor_labels: true
honor_timestamps: true
scheme: https
params:
foo:
- x
authorization:
type: foobar
headers:
- 'TenantID: fooBar'
- 'X: y:z'
relabel_configs:
- source_labels: [abc]
static_configs:
- targets:
- foo
scrape_align_interval: 1h30m0s
proxy_bearer_token_file: file.txt
proxy_headers:
- 'My-Auth-Header: top-secret'
`)
}
func TestGetClusterMemberNumsForScrapeWork(t *testing.T) {
f := func(key string, membersCount, replicationFactor int, expectedMemberNums []int) {
t.Helper()
memberNums := getClusterMemberNumsForScrapeWork(key, membersCount, replicationFactor)
if !reflect.DeepEqual(memberNums, expectedMemberNums) {
t.Fatalf("unexpected memberNums; got %d; want %d", memberNums, expectedMemberNums)
}
}
// Disabled clustering
f("foo", 0, 0, []int{0})
f("foo", 0, 0, []int{0})
// A cluster with 2 nodes with disabled replication
f("baz", 2, 0, []int{0})
f("foo", 2, 0, []int{1})
// A cluster with 2 nodes with replicationFactor=2
f("baz", 2, 2, []int{0, 1})
f("foo", 2, 2, []int{1, 0})
// A cluster with 3 nodes with replicationFactor=2
f("abc", 3, 2, []int{0, 1})
f("bar", 3, 2, []int{1, 2})
f("foo", 3, 2, []int{2, 0})
}
func TestLoadStaticConfigs(t *testing.T) {
scs, err := loadStaticConfigs("testdata/file_sd.json")
if err != nil {
t.Fatalf("unexpected error: %s", err)
}
if len(scs) == 0 {
t.Fatalf("expecting non-zero static configs")
}
// Try loading non-existing file
scs, err = loadStaticConfigs("testdata/non-exsiting-file")
if err == nil {
t.Fatalf("expecting non-nil error")
}
if scs != nil {
t.Fatalf("unexpected non-nil static configs: %#v", scs)
}
// Try loading invalid file
scs, err = loadStaticConfigs("testdata/prometheus.yml")
if err == nil {
t.Fatalf("expecting non-nil error")
}
if scs != nil {
t.Fatalf("unexpected non-nil static configs: %#v", scs)
}
}
func TestLoadConfig(t *testing.T) {
cfg, err := loadConfig("testdata/prometheus.yml")
if err != nil {
t.Fatalf("unexpected error: %s", err)
}
if cfg == nil {
t.Fatalf("expecting non-nil config")
}
cfg, err = loadConfig("testdata/prometheus-with-scrape-config-files.yml")
if err != nil {
t.Fatalf("unexpected error: %s", err)
}
if cfg == nil {
t.Fatalf("expecting non-nil config")
}
// Try loading non-existing file
cfg, err = loadConfig("testdata/non-existing-file")
if err == nil {
t.Fatalf("expecting non-nil error")
}
if cfg != nil {
t.Fatalf("unexpected non-nil config: %#v", cfg)
}
// Try loading invalid file
cfg, err = loadConfig("testdata/file_sd_1.yml")
if err == nil {
t.Fatalf("expecting non-nil error")
}
if cfg != nil {
t.Fatalf("unexpected non-nil config: %#v", cfg)
}
}
func TestAddressWithFullURL(t *testing.T) {
data := `
scrape_configs:
- job_name: abc
metrics_path: /foo/bar
scheme: https
params:
x: [y]
static_configs:
- targets:
# the following targets are scraped by the provided urls
- 'http://host1/metric/path1'
- 'https://host2/metric/path2'
- 'http://host3:1234/metric/path3?arg1=value1'
# the following target is scraped by <scheme>://host4:1234<metrics_path>
- host4:1234
`
var cfg Config
if err := cfg.parseData([]byte(data), "sss"); err != nil {
t.Fatalf("cannot parase data: %s", err)
}
sws := cfg.getStaticScrapeWork()
swsExpected := []*ScrapeWork{
{
ScrapeURL: "http://host1/metric/path1?x=y",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "host1:80",
"job": "abc",
}),
jobNameOriginal: "abc",
},
{
ScrapeURL: "https://host2/metric/path2?x=y",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "host2:443",
"job": "abc",
}),
jobNameOriginal: "abc",
},
{
ScrapeURL: "http://host3:1234/metric/path3?arg1=value1&x=y",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "host3:1234",
"job": "abc",
}),
jobNameOriginal: "abc",
},
{
ScrapeURL: "https://host4:1234/foo/bar?x=y",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "host4:1234",
"job": "abc",
}),
jobNameOriginal: "abc",
},
}
checkEqualScrapeWorks(t, sws, swsExpected)
}
func TestBlackboxExporter(t *testing.T) {
// See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/684
data := `
scrape_configs:
- job_name: 'blackbox'
metrics_path: /probe
params:
module: [dns_udp_example] # Look for dns response
static_configs:
- targets:
- 8.8.8.8
relabel_configs:
- source_labels: [__address__]
target_label: __param_target
- source_labels: [__param_target]
target_label: instance
- target_label: __address__
replacement: black:9115 # The blackbox exporter's real hostname:port.%
`
var cfg Config
if err := cfg.parseData([]byte(data), "sss"); err != nil {
t.Fatalf("cannot parase data: %s", err)
}
sws := cfg.getStaticScrapeWork()
swsExpected := []*ScrapeWork{{
ScrapeURL: "http://black:9115/probe?module=dns_udp_example&target=8.8.8.8",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "8.8.8.8",
"job": "blackbox",
}),
jobNameOriginal: "blackbox",
}}
checkEqualScrapeWorks(t, sws, swsExpected)
}
func TestGetFileSDScrapeWork(t *testing.T) {
data := `
scrape_configs:
- job_name: foo
file_sd_configs:
- files: [testdata/file_sd.json]
`
var cfg Config
if err := cfg.parseData([]byte(data), "sss"); err != nil {
t.Fatalf("cannot parase data: %s", err)
}
sws := cfg.getFileSDScrapeWork(nil)
if !equalStaticConfigForScrapeWorks(sws, sws) {
t.Fatalf("unexpected non-equal static configs;\nsws:\n%#v", sws)
}
// Load another static config
dataNew := `
scrape_configs:
- job_name: foo
file_sd_configs:
- files: [testdata/file_sd_1.yml]
`
var cfgNew Config
if err := cfgNew.parseData([]byte(dataNew), "sss"); err != nil {
t.Fatalf("cannot parse data: %s", err)
}
swsNew := cfgNew.getFileSDScrapeWork(sws)
if equalStaticConfigForScrapeWorks(swsNew, sws) {
t.Fatalf("unexpected equal static configs;\nswsNew:\n%#v\nsws:\n%#v", swsNew, sws)
}
// Try loading invalid static config
data = `
scrape_configs:
- job_name: foo
file_sd_configs:
- files: [testdata/prometheus.yml]
`
if err := cfg.parseData([]byte(data), "sss"); err != nil {
t.Fatalf("cannot parse data: %s", err)
}
sws = cfg.getFileSDScrapeWork(swsNew)
if len(sws) != 0 {
t.Fatalf("unexpected non-empty sws:\n%#v", sws)
}
// Empty target in static config
data = `
scrape_configs:
- job_name: foo
file_sd_configs:
- files: [testdata/empty_target_file_sd.yml]
`
if err := cfg.parseData([]byte(data), "sss"); err != nil {
t.Fatalf("cannot parse data: %s", err)
}
sws = cfg.getFileSDScrapeWork(swsNew)
if len(sws) != 0 {
t.Fatalf("unexpected non-empty sws:\n%#v", sws)
}
}
func getFileSDScrapeWork(data []byte, path string) ([]*ScrapeWork, error) {
var cfg Config
if err := cfg.parseData(data, path); err != nil {
return nil, fmt.Errorf("cannot parse data: %w", err)
}
return cfg.getFileSDScrapeWork(nil), nil
}
func getStaticScrapeWork(data []byte, path string) ([]*ScrapeWork, error) {
var cfg Config
if err := cfg.parseData(data, path); err != nil {
return nil, fmt.Errorf("cannot parse data: %w", err)
}
return cfg.getStaticScrapeWork(), nil
}
func TestGetStaticScrapeWorkFailure(t *testing.T) {
f := func(data string) {
t.Helper()
sws, err := getStaticScrapeWork([]byte(data), "non-existing-file")
if err == nil {
t.Fatalf("expecting non-nil error")
}
if sws != nil {
t.Fatalf("expecting nil sws")
}
}
// incorrect yaml
f(`foo bar baz`)
// yaml with unsupported fields
f(`foo: bar`)
f(`
scrape_configs:
- foo: bar
`)
// invalid scrape_config_files contents
f(`
scrape_config_files:
- job_name: aa
static_configs:
- targets: ["s"]
`)
// Duplicate job_name
f(`
scrape_configs:
- job_name: foo
static_configs:
targets: ["foo"]
- job_name: foo
static_configs:
targets: ["bar"]
`)
}
// String returns human-readable representation for sw.
func (sw *ScrapeWork) String() string {
return stringsutil.JSONString(sw.key())
}
func TestGetFileSDScrapeWorkSuccess(t *testing.T) {
f := func(data string, expectedSws []*ScrapeWork) {
t.Helper()
sws, err := getFileSDScrapeWork([]byte(data), "non-existing-file")
if err != nil {
t.Fatalf("unexpected error: %s", err)
}
checkEqualScrapeWorks(t, sws, expectedSws)
}
f(`
scrape_configs:
- job_name: foo
static_configs:
- targets: ["xxx"]
`, []*ScrapeWork{})
f(`
scrape_configs:
- job_name: foo
metrics_path: /abc/de
file_sd_configs:
- files: ["testdata/file_sd.json", "testdata/file_sd*.yml"]
`, []*ScrapeWork{
{
ScrapeURL: "http://host1/abc/de",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "host1:80",
"job": "foo",
"qwe": "rty",
}),
jobNameOriginal: "foo",
},
{
ScrapeURL: "http://host2/abc/de",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "host2:80",
"job": "foo",
"qwe": "rty",
}),
jobNameOriginal: "foo",
},
{
ScrapeURL: "http://localhost:9090/abc/de",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "localhost:9090",
"job": "foo",
"yml": "test",
}),
jobNameOriginal: "foo",
},
})
}
func TestGetStaticScrapeWorkSuccess(t *testing.T) {
f := func(data string, expectedSws []*ScrapeWork) {
t.Helper()
sws, err := getStaticScrapeWork([]byte(data), "non-exsiting-file")
if err != nil {
t.Fatalf("unexpected error: %s", err)
}
checkEqualScrapeWorks(t, sws, expectedSws)
}
f(``, nil)
// Scrape config with missing modulus for action=hashmod in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- action: hashmod
source_labels: [foo]
target_label: bar
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with invalid action in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- action: foobar
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with missing source_labels for action=keep in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- action: keep
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with missing source_labels for action=drop in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- action: drop
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with missing source_labels for action=hashmod in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- action: hashmod
target_label: bar
modulus: 123
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with missing target for action=hashmod in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- action: hashmod
source_labels: [foo]
modulus: 123
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with invalid regex in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- regex: "("
source_labels: [foo]
target_label: bar
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with missing target_label for action=replace in relabel_configs must be skipped
f(`
scrape_configs:
- job_name: aa
relabel_configs:
- action: replace
source_labels: [foo]
static_configs:
- targets: ["s"]
`, []*ScrapeWork{})
// Scrape config with both `authorization` and `bearer_token` set must be skipped
f(`
scrape_configs:
- job_name: x
authorization:
credentials: foobar
bearer_token: foo
static_configs:
- targets: ["a"]
`, []*ScrapeWork{})
// Scrape config with both `bearer_token` and `bearer_token_file` set must be skipped
f(`
scrape_configs:
- job_name: x
bearer_token: foo
bearer_token_file: bar
static_configs:
- targets: ["a"]
`, []*ScrapeWork{})
// Scrape config with both `basic_auth` and `bearer_token` set must be skipped
f(`
scrape_configs:
- job_name: x
bearer_token: foo
basic_auth:
username: foo
password: bar
static_configs:
- targets: ["a"]
`, []*ScrapeWork{})
// Scrape config with both `authorization` and `basic_auth` set must be skipped
f(`
scrape_configs:
- job_name: x
authorization:
credentials: foobar
basic_auth:
username: foobar
static_configs:
- targets: ["a"]
`, []*ScrapeWork{})
// Scrape config with invalid scheme must be skipped
f(`
scrape_configs:
- job_name: x
scheme: asdf
static_configs:
- targets: ["foo"]
`, []*ScrapeWork{})
// Scrape config with missing job_name must be skipped
f(`
scrape_configs:
- static_configs:
- targets: ["foo"]
`, []*ScrapeWork{})
// Scrape config with missing username in `basic_auth` must be skipped
f(`
scrape_configs:
- job_name: x
basic_auth:
password: sss
static_configs:
- targets: ["a"]
`, []*ScrapeWork{})
// Scrape config with both password and password_file set in `basic_auth` must be skipped
f(`
scrape_configs:
- job_name: x
basic_auth:
username: foobar
password: sss
password_file: sdfdf
static_configs:
- targets: ["a"]
`, []*ScrapeWork{})
// Scrape config with invalid ca_file must be properly parsed, since ca_file may become valid later
f(`
scrape_configs:
- job_name: aa
tls_config:
ca_file: testdata/prometheus.yml
static_configs:
- targets: ["s"]
`, []*ScrapeWork{
{
ScrapeURL: "http://s/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "s:80",
"job": "aa",
}),
jobNameOriginal: "aa",
},
})
// Scrape config with non-existing ca_file must be properly parsed, since the ca_file can become valid later
f(`
scrape_configs:
- job_name: aa
tls_config:
ca_file: non/extising/file
static_configs:
- targets: ["s"]
`, []*ScrapeWork{
{
ScrapeURL: "http://s/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "s:80",
"job": "aa",
}),
jobNameOriginal: "aa",
},
})
// Scrape config with non-existing cert_file must be properly parsed, since the cert_file can become valid later
f(`
scrape_configs:
- job_name: aa
tls_config:
cert_file: non/extising/file
static_configs:
- targets: ["s"]
`, []*ScrapeWork{
{
ScrapeURL: "http://s/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "s:80",
"job": "aa",
}),
jobNameOriginal: "aa",
},
})
// Scrape config with non-existing key_file must be properly parsed, since the key_file can become valid later
f(`
scrape_configs:
- job_name: aa
tls_config:
key_file: non/extising/file
static_configs:
- targets: ["s"]
`, []*ScrapeWork{
{
ScrapeURL: "http://s/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "s:80",
"job": "aa",
}),
jobNameOriginal: "aa",
},
})
f(`
scrape_configs:
- job_name: foo
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "foo",
}),
jobNameOriginal: "foo",
},
})
f(`
global:
external_labels:
datacenter: foobar
jobs: xxx
scrape_configs:
- job_name: foo
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "foo",
}),
ExternalLabels: promutils.NewLabelsFromMap(map[string]string{
"datacenter": "foobar",
"jobs": "xxx",
}),
jobNameOriginal: "foo",
},
})
f(`
global:
scrape_interval: 8s
scrape_timeout: 34s
scrape_configs:
- job_name: foo
scrape_interval: 54s
scrape_timeout: 12s
metrics_path: /foo/bar
scheme: https
honor_labels: true
honor_timestamps: true
follow_redirects: false
params:
p: ["x&y", "="]
xaa:
proxy_url: http://foo.bar
static_configs:
- targets: ["foo.bar", "aaa"]
labels:
x: y
__scrape_timeout__: "5s"
- job_name: qwer
tls_config:
server_name: foobar
insecure_skip_verify: true
static_configs:
- targets: [1.2.3.4]
- job_name: asdf
static_configs:
- targets: [foobar]
`, []*ScrapeWork{
{
ScrapeURL: "https://foo.bar/foo/bar?p=x%26y&p=%3D",
ScrapeInterval: 54 * time.Second,
ScrapeTimeout: 5 * time.Second,
MaxScrapeSize: maxScrapeSize.N,
HonorLabels: true,
HonorTimestamps: true,
DenyRedirects: true,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:443",
"job": "foo",
"x": "y",
}),
ProxyURL: proxy.MustNewURL("http://foo.bar"),
jobNameOriginal: "foo",
},
{
ScrapeURL: "https://aaa/foo/bar?p=x%26y&p=%3D",
ScrapeInterval: 54 * time.Second,
ScrapeTimeout: 5 * time.Second,
MaxScrapeSize: maxScrapeSize.N,
HonorLabels: true,
HonorTimestamps: true,
DenyRedirects: true,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "aaa:443",
"job": "foo",
"x": "y",
}),
ProxyURL: proxy.MustNewURL("http://foo.bar"),
jobNameOriginal: "foo",
},
{
ScrapeURL: "http://1.2.3.4/metrics",
ScrapeInterval: 8 * time.Second,
ScrapeTimeout: 8 * time.Second,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "1.2.3.4:80",
"job": "qwer",
}),
jobNameOriginal: "qwer",
},
{
ScrapeURL: "http://foobar/metrics",
ScrapeInterval: 8 * time.Second,
ScrapeTimeout: 8 * time.Second,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foobar:80",
"job": "asdf",
}),
jobNameOriginal: "asdf",
},
})
f(`
scrape_configs:
- job_name: foo
relabel_configs:
- source_labels: [__scheme__, __address__]
separator: "://"
target_label: __tmp_url
- source_labels: [__tmp_url, __metrics_path__]
separator: ""
target_label: url
- action: labeldrop
regex: "job|__tmp_.+"
- action: drop
source_labels: [__address__]
regex: "drop-.*"
- action: keep
source_labels: [__param_x]
regex: keep_me
- action: labelkeep
regex: "__.*|url"
- action: labelmap
regex: "(url)"
replacement: "prefix:${1}"
- action: hashmod
modulus: 123
source_labels: [__address__]
target_label: hash
- action: replace
source_labels: [__address__]
target_label: foobar
replacement: ""
params:
x: [keep_me]
static_configs:
- targets: ["foo.bar:1234", "drop-this-target"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics?x=keep_me",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"hash": "82",
"instance": "foo.bar:1234",
"prefix:url": "http://foo.bar:1234/metrics",
"url": "http://foo.bar:1234/metrics",
}),
jobNameOriginal: "foo",
},
})
f(`
scrape_configs:
- job_name: foo
scheme: https
relabel_configs:
- action: replace
source_labels: [non-existing-label]
target_label: instance
replacement: fake.addr
- action: replace
source_labels: [__address__]
target_label: foobar
regex: "missing-regex"
replacement: aaabbb
- action: replace
source_labels: [__scheme__]
target_label: job
- action: replace
source_labels: [__scheme__]
target_label: __scheme__
replacement: mailto
- target_label: __metrics_path__
replacement: /abc.de
- target_label: __param_a
replacement: b
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "mailto://foo.bar:1234/abc.de?a=b",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "fake.addr",
"job": "https",
}),
jobNameOriginal: "foo",
},
})
f(`
scrape_configs:
- job_name: foo
scheme: https
max_scrape_size: 1
relabel_configs:
- action: keep
source_labels: [__address__]
regex: "foo\\.bar:.*"
- action: hashmod
source_labels: [job]
modulus: 4
target_label: job
- action: labeldrop
regex: "non-matching-regex"
- action: labelkeep
regex: "job|__address__"
- action: labeldrop
regex: ""
static_configs:
- targets: ["foo.bar:1234", "xyz"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: 1,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "3",
}),
jobNameOriginal: "foo",
},
})
f(`
scrape_configs:
- job_name: foo
max_scrape_size: 8MiB
metric_relabel_configs:
- source_labels: [foo]
target_label: abc
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: 8 * 1024 * 1024,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "foo",
}),
jobNameOriginal: "foo",
},
})
f(`
scrape_configs:
- job_name: foo
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "foo",
}),
jobNameOriginal: "foo",
},
})
f(`
scrape_configs:
- job_name: foo
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "foo",
}),
jobNameOriginal: "foo",
},
})
f(`
global:
external_labels:
job: foobar
foo: xx
q: qwe
__address__: aaasdf
__param_a: jlfd
scrape_configs:
- job_name: aaa
params:
a: [b, xy]
static_configs:
- targets: ["a"]
labels:
foo: bar
__param_a: c
__address__: pp
job: yyy
`, []*ScrapeWork{
{
ScrapeURL: "http://pp/metrics?a=c&a=xy",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"foo": "bar",
"instance": "pp:80",
"job": "yyy",
}),
ExternalLabels: promutils.NewLabelsFromMap(map[string]string{
"__address__": "aaasdf",
"__param_a": "jlfd",
"foo": "xx",
"job": "foobar",
"q": "qwe",
}),
jobNameOriginal: "aaa",
},
})
f(`
scrape_configs:
- job_name: 'snmp'
sample_limit: 100
disable_keepalive: true
disable_compression: true
headers:
- "My-Auth: foo-Bar"
proxy_headers:
- "Foo: bar"
scrape_align_interval: 1s
scrape_offset: 0.5s
static_configs:
- targets:
- 192.168.1.2 # SNMP device.
metrics_path: /snmp
params:
module: [if_mib]
relabel_configs:
- source_labels: [__address__]
target_label: __param_target
- source_labels: [__param_target]
target_label: instance
- target_label: __address__
replacement: 127.0.0.1:9116 # The SNMP exporter's real hostname:port.
- target_label: __series_limit__
replacement: 1234
- target_label: __sample_limit__
replacement: 5678
- target_label: __stream_parse__
replacement: true
`, []*ScrapeWork{
{
ScrapeURL: "http://127.0.0.1:9116/snmp?module=if_mib&target=192.168.1.2",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "192.168.1.2",
"job": "snmp",
}),
SampleLimit: 5678,
DisableKeepAlive: true,
DisableCompression: true,
StreamParse: true,
ScrapeAlignInterval: time.Second,
ScrapeOffset: 500 * time.Millisecond,
SeriesLimit: 1234,
jobNameOriginal: "snmp",
},
})
f(`
scrape_configs:
- job_name: path wo slash
enable_compression: false
static_configs:
- targets: ["foo.bar:1234"]
relabel_configs:
- replacement: metricspath
target_label: __metrics_path__
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metricspath",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "path wo slash",
}),
DisableCompression: true,
jobNameOriginal: "path wo slash",
},
})
f(`
global:
scrape_timeout: 1d
scrape_configs:
- job_name: foo
scrape_interval: 1w
scrape_align_interval: 1d
scrape_offset: 2d
no_stale_markers: true
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: time.Hour * 24 * 7,
ScrapeTimeout: time.Hour * 24,
ScrapeAlignInterval: time.Hour * 24,
ScrapeOffset: time.Hour * 24 * 2,
MaxScrapeSize: maxScrapeSize.N,
NoStaleMarkers: true,
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "foo",
}),
jobNameOriginal: "foo",
},
})
defaultSeriesLimitPerTarget := *seriesLimitPerTarget
*seriesLimitPerTarget = 1e3
f(`
scrape_configs:
- job_name: foo
series_limit: 0
static_configs:
- targets: ["foo.bar:1234"]
`, []*ScrapeWork{
{
ScrapeURL: "http://foo.bar:1234/metrics",
ScrapeInterval: defaultScrapeInterval,
ScrapeTimeout: defaultScrapeTimeout,
MaxScrapeSize: maxScrapeSize.N,
jobNameOriginal: "foo",
Labels: promutils.NewLabelsFromMap(map[string]string{
"instance": "foo.bar:1234",
"job": "foo",
}),
SeriesLimit: 0,
},
})
*seriesLimitPerTarget = defaultSeriesLimitPerTarget
}
func equalStaticConfigForScrapeWorks(a, b []*ScrapeWork) bool {
if len(a) != len(b) {
return false
}
for i := range a {
keyA := a[i].key()
keyB := b[i].key()
if keyA != keyB {
return false
}
}
return true
}
func TestScrapeConfigClone(t *testing.T) {
f := func(sc *ScrapeConfig) {
t.Helper()
scCopy := sc.clone()
scJSON := sc.marshalJSON()
scCopyJSON := scCopy.marshalJSON()
if !reflect.DeepEqual(scJSON, scCopyJSON) {
t.Fatalf("unexpected cloned result:\ngot\n%s\nwant\n%s", scCopyJSON, scJSON)
}
}
f(&ScrapeConfig{})
var ie promrelabel.IfExpression
if err := ie.Parse(`{foo=~"bar",baz!="z"}`); err != nil {
t.Fatalf("unexpected error: %s", err)
}
f(&ScrapeConfig{
JobName: "foo",
ScrapeInterval: promutils.NewDuration(time.Second * 47),
HonorLabels: true,
Params: map[string][]string{
"foo": {"bar", "baz"},
},
HTTPClientConfig: promauth.HTTPClientConfig{
Authorization: &promauth.Authorization{
Credentials: promauth.NewSecret("foo"),
},
BasicAuth: &promauth.BasicAuthConfig{
Username: "user_x",
Password: promauth.NewSecret("pass_x"),
},
BearerToken: promauth.NewSecret("zx"),
OAuth2: &promauth.OAuth2Config{
ClientSecret: promauth.NewSecret("aa"),
Scopes: []string{"foo", "bar"},
TLSConfig: &promauth.TLSConfig{
CertFile: "foo",
},
},
TLSConfig: &promauth.TLSConfig{
KeyFile: "aaa",
},
},
ProxyURL: proxy.MustNewURL("https://foo.bar:3434/assdf/dsfd?sdf=dsf"),
RelabelConfigs: []promrelabel.RelabelConfig{{
SourceLabels: []string{"foo", "aaa"},
Regex: &promrelabel.MultiLineRegex{
S: "foo\nbar",
},
If: &ie,
}},
SampleLimit: 10,
GCESDConfigs: []gce.SDConfig{{
Project: "foo",
Zone: gce.ZoneYAML{
Zones: []string{"a", "b"},
},
}},
StreamParse: true,
ProxyClientConfig: promauth.ProxyClientConfig{
BearerTokenFile: "foo",
},
})
}
func checkEqualScrapeWorks(t *testing.T, got, want []*ScrapeWork) {
t.Helper()
if len(got) != len(want) {
t.Fatalf("unexpected number of ScrapeWork items; got %d; want %d", len(got), len(want))
}
for i := range got {
gotItem := *got[i]
wantItem := want[i]
// Zero fields with internal state before comparing the items.
gotItem.ProxyAuthConfig = nil
gotItem.AuthConfig = nil
gotItem.OriginalLabels = nil
gotItem.RelabelConfigs = nil
gotItem.MetricRelabelConfigs = nil
if !reflect.DeepEqual(&gotItem, wantItem) {
t.Fatalf("unexpected scrapeWork at position %d out of %d;\ngot\n%#v\nwant\n%#v", i, len(got), &gotItem, wantItem)
}
}
}