Generate default client IDs using host and client start time (#255)

Here, move away from ULIDs for generating client IDs, and over to a
custom ID generator that uses a combination of host and client start
time. This has two benefits:

* Host/start time are more descriptive for humans. It should give an
  operator a rough idea as to where the name of a client originated
  from, and if it's a client that's likely long dead (if the timestamp
  is very old).

* Lets us drop our ULID package dependency, which we don't use for
  anything else.

Generated IDs look like this for example:

    miniml2_local_2024_03_07T04_39_12

I've tried to arrange them so that on most systems you'll be able to
double click on an ID to select the whole thing, which might be helpful
while copying information or debugging a production problem. Characters
like colons (`:`) are less friendly for double-click-to-select, so I've
avoided them. Hyphens and dots aren't universally friendly for double
click either (depends on program and configuration), so I've avoided
those as well.

CHANGELOG.md

@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Changed
+
+- Changed default client IDs to be a combination of hostname and the time which the client started. This can still be changed by specifying `Config.ID`. [PR #255](https://github.com/riverqueue/river/pull/255).
+
 ## [0.0.25] - 2024-03-01
 
 ### Fixed

client.go

@@ -2,18 +2,16 @@ package river
 
 import (
 	"context"
-	"crypto/rand"
 	"encoding/json"
 	"errors"
 	"fmt"
 	"log/slog"
 	"os"
 	"regexp"
+	"strings"
 	"sync"
 	"time"
 
-	"github.com/oklog/ulid/v2"
-
 	"github.com/riverqueue/river/internal/baseservice"
 	"github.com/riverqueue/river/internal/componentstatus"
 	"github.com/riverqueue/river/internal/dblist"
@@ -405,7 +403,7 @@ func NewClient[TTx any](driver riverdriver.Driver[TTx], config *Config) (*Client
 		ErrorHandler:                config.ErrorHandler,
 		FetchCooldown:               valutil.ValOrDefault(config.FetchCooldown, FetchCooldownDefault),
 		FetchPollInterval:           valutil.ValOrDefault(config.FetchPollInterval, FetchPollIntervalDefault),
-		ID:                          config.ID,
+		ID:                          valutil.ValOrDefaultFunc(config.ID, func() string { return defaultClientID(time.Now().UTC()) }),
 		JobTimeout:                  valutil.ValOrDefault(config.JobTimeout, JobTimeoutDefault),
 		Logger:                      logger,
 		PeriodicJobs:                config.PeriodicJobs,
@@ -418,15 +416,6 @@ func NewClient[TTx any](driver riverdriver.Driver[TTx], config *Config) (*Client
 		schedulerInterval:           valutil.ValOrDefault(config.schedulerInterval, maintenance.JobSchedulerIntervalDefault),
 	}
 
-	if config.ID == "" {
-		// Generate a random ULID for the client ID.
-		clientID, err := ulid.New(ulid.Now(), rand.Reader)
-		if err != nil {
-			return nil, err
-		}
-		config.ID = clientID.String()
-	}
-
 	if err := config.validate(); err != nil {
 		return nil, err
 	}
@@ -1398,3 +1387,31 @@ func (c *Client[TTx]) JobListTx(ctx context.Context, tx TTx, params *JobListPara
 
 	return dblist.JobList(ctx, c.driver.UnwrapExecutor(tx), dbParams)
 }
+
+// Generates a default client ID using the current hostname and time.
+func defaultClientID(startedAt time.Time) string {
+	host, _ := os.Hostname()
+	if host == "" {
+		host = "unknown_host"
+	}
+
+	return defaultClientIDWithHost(startedAt, host)
+}
+
+// Same as the above, but allows host injection for testability.
+func defaultClientIDWithHost(startedAt time.Time, host string) string {
+	const maxHostLength = 60
+
+	// Truncate degenerately long host names.
+	host = strings.ReplaceAll(host, ".", "_")
+	if len(host) > maxHostLength {
+		host = host[0:maxHostLength]
+	}
+
+	// Dots, hyphens, and colons aren't particularly friendly for double click
+	// to select (depends on application and configuration), so avoid them all
+	// in favor of underscores.
+	const rfc3339Compact = "2006_01_02T15_04_05"
+
+	return host + "_" + startedAt.Format(rfc3339Compact)
+}

client_test.go

@@ -3600,3 +3600,33 @@ func TestUniqueOpts(t *testing.T) {
 		require.NotEqual(t, job0.ID, job2.ID)
 	})
 }
+
+func TestDefaultClientID(t *testing.T) {
+	t.Parallel()
+
+	host, _ := os.Hostname()
+	require.NotEmpty(t, host)
+
+	startedAt := time.Date(2024, time.March, 7, 4, 39, 12, 0, time.UTC)
+
+	require.Equal(t, strings.ReplaceAll(host, ".", "_")+"_2024_03_07T04_39_12", defaultClientID(startedAt)) //nolint:goconst
+}
+
+func TestDefaultClientIDWithHost(t *testing.T) {
+	t.Parallel()
+
+	host, _ := os.Hostname()
+	require.NotEmpty(t, host)
+
+	startedAt := time.Date(2024, time.March, 7, 4, 39, 12, 0, time.UTC)
+
+	require.Equal(t, "example_com_2024_03_07T04_39_12", defaultClientIDWithHost(startedAt,
+		"example.com"))
+	require.Equal(t, "this_is_a_degenerately_long_host_name_that_will_be_truncated_2024_03_07T04_39_12", defaultClientIDWithHost(startedAt,
+		"this.is.a.degenerately.long.host.name.that.will.be.truncated.so.were.not.storing.massive.strings.to.the.database.com"))
+
+	// Test strings right around the boundary to make sure we don't have some off-by-one slice error.
+	require.Equal(t, strings.Repeat("a", 59)+"_2024_03_07T04_39_12", defaultClientIDWithHost(startedAt, strings.Repeat("a", 59)))
+	require.Equal(t, strings.Repeat("a", 60)+"_2024_03_07T04_39_12", defaultClientIDWithHost(startedAt, strings.Repeat("a", 60)))
+	require.Equal(t, strings.Repeat("a", 60)+"_2024_03_07T04_39_12", defaultClientIDWithHost(startedAt, strings.Repeat("a", 61)))
+}

go.mod

@@ -14,7 +14,6 @@ require (
 	github.com/jackc/pgerrcode v0.0.0-20220416144525-469b46aa5efa
 	github.com/jackc/pgx/v5 v5.5.3
 	github.com/jackc/puddle/v2 v2.2.1
-	github.com/oklog/ulid/v2 v2.1.0
 	github.com/riverqueue/river/riverdriver v0.0.25
 	github.com/riverqueue/river/riverdriver/riverdatabasesql v0.0.25
 	github.com/riverqueue/river/riverdriver/riverpgxv5 v0.0.25

go.sum

@@ -20,9 +20,6 @@ github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
 github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
 github.com/lib/pq v1.10.9 h1:YXG7RB+JIjhP29X+OtkiDnYaXQwpS4JEWq7dtCCRUEw=
 github.com/lib/pq v1.10.9/go.mod h1:AlVN5x4E4T544tWzH6hKfbfQvm3HdbOxrmggDNAPY9o=
-github.com/oklog/ulid/v2 v2.1.0 h1:+9lhoxAP56we25tyYETBBY1YLA2SaoLvUFgrP2miPJU=
-github.com/oklog/ulid/v2 v2.1.0/go.mod h1:rcEKHmBBKfef9DhnvX7y1HZBYxjXb0cP5ExxNsTT1QQ=
-github.com/pborman/getopt v0.0.0-20170112200414-7148bc3a4c30/go.mod h1:85jBQOZwpVEaDAr341tbn15RS4fCAsIst0qp7i8ex1o=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/robfig/cron/v3 v3.0.1 h1:WdRxkvbJztn8LMz/QEvLN5sBU+xKpSqwwUO1Pjr4qDs=

internal/util/valutil/val_util.go

@@ -12,6 +12,16 @@ func ValOrDefault[T constraints.Integer | string](val, defaultVal T) T {
 	return defaultVal
 }
 
+// ValOrDefault returns the given value if it's non-zero, and otherwise invokes
+// defaultFunc to produce a default value.
+func ValOrDefaultFunc[T constraints.Integer | string](val T, defaultFunc func() T) T {
+	var zero T
+	if val != zero {
+		return val
+	}
+	return defaultFunc()
+}
+
 // FirstNonZero returns the first argument that is non-zero, or the zero value if
 // all are zero.
 func FirstNonZero[T constraints.Integer | ~string](values ...T) T {

internal/util/valutil/val_util_test.go

@@ -15,3 +15,13 @@ func TestValOrDefault(t *testing.T) {
 	require.Equal(t, "default", ValOrDefault("", "default"))
 	require.Equal(t, "hello", ValOrDefault("hello", "default"))
 }
+
+func TestValOrDefaultFunc(t *testing.T) {
+	t.Parallel()
+
+	require.Equal(t, 1, ValOrDefaultFunc(0, func() int { return 1 }))
+	require.Equal(t, 5, ValOrDefaultFunc(5, func() int { return 1 }))
+
+	require.Equal(t, "default", ValOrDefaultFunc("", func() string { return "default" }))
+	require.Equal(t, "hello", ValOrDefaultFunc("hello", func() string { return "default" }))
+}