feat: Add "queryMaxLength" config var

Before this the 10000 length truncation of the "query" field for
database spans was hardcoded.

Fixes: #1921

docs/configuration.asciidoc

@@ -689,6 +689,25 @@ This applies to the following properties:
 - `error.exception.message`
 - `error.log.message`
 
+
+[[query-max-length]]
+==== `queryMaxLength`
+
+* *Type:* String
+* *Default:* `10000b`
+* *Env:* `ELASTIC_APM_QUERY_MAX_LENGTH`
+
+The query captured for database spans will be truncated at this length.
+It is expressed as an integer size and a unit (one of `b`, `kb`, `mb`, `gb`,
+case-insensitive).  For example, `100b` is 100 bytes, `2kb` is 2048 bytes, `1mb`
+is 1048576 bytes.
+
+Note that span data is limited at the upstream APM server to
+{apm-server-ref-v}/configuration-process.html#max_event_size[`max_event_size`],
+which defaults to 300kB. If you configure `queryMaxLength` too large, it could
+result in spans that are rejected by APM server.
+
+
 [[stack-trace-limit]]
 ==== `stackTraceLimit`
 

index.d.ts

@@ -233,6 +233,7 @@ interface AgentConfigOptions {
   metricsLimit?: number;
   payloadLogFile?: string;
   centralConfig?: boolean;
+  queryMaxLength?: string;
   sanitizeFieldNames?: Array<string>;
   secretToken?: string;
   serverCaCertFile?: string;

lib/config.js

@@ -75,6 +75,7 @@ var DEFAULTS = {
   maxQueueSize: 1024,
   metricsInterval: '30s',
   metricsLimit: 1000,
+  queryMaxLength: '10000b',
   sanitizeFieldNames: ['password', 'passwd', 'pwd', 'secret', '*key', '*token*',
     '*session*', '*credit*', '*card*', 'authorization', 'set-cookie',
     'pw', 'pass', 'connect.sid'
@@ -135,6 +136,7 @@ var ENV_TABLE = {
   metricsInterval: 'ELASTIC_APM_METRICS_INTERVAL',
   metricsLimit: 'ELASTIC_APM_METRICS_LIMIT',
   payloadLogFile: 'ELASTIC_APM_PAYLOAD_LOG_FILE',
+  queryMaxLength: 'ELASTIC_APM_QUERY_MAX_LENGTH',
   sanitizeFieldNames: 'ELASTIC_SANITIZE_FIELD_NAMES',
   serverCaCertFile: 'ELASTIC_APM_SERVER_CA_CERT_FILE',
   secretToken: 'ELASTIC_APM_SECRET_TOKEN',
@@ -205,10 +207,16 @@ var TIME_OPTS = [
   'spanFramesMinDuration'
 ]
 
+// All new config vars should use STRICT_BYTES_OPTS rather than BYTES_OPTS
+// to require the unit.
+// https://github.com/elastic/apm/blob/master/specs/agents/configuration.md#durationsize-config-legacy-considerations
 var BYTES_OPTS = [
   'apiRequestSize',
   'errorMessageMaxLength'
 ]
+var STRICT_BYTES_OPTS = [
+  'queryMaxLength'
+]
 
 var MINUS_ONE_EQUAL_INFINITY = [
   'transactionMaxSpans'
@@ -399,6 +407,7 @@ function normalize (opts, logger) {
   normalizeKeyValuePairs(opts)
   normalizeNumbers(opts)
   normalizeBytes(opts)
+  normalizeStrictBytes(opts, logger)
   normalizeArrays(opts)
   normalizeTime(opts)
   normalizeBools(opts, logger)
@@ -508,7 +517,29 @@ function normalizeNumbers (opts) {
 
 function normalizeBytes (opts) {
   for (const key of BYTES_OPTS) {
-    if (key in opts) opts[key] = bytes(String(opts[key]))
+    if (key in opts) {
+      opts[key] = numBytesFromStr(String(opts[key]), false)
+    }
+  }
+}
+
+function normalizeStrictBytes (opts, logger) {
+  for (const key of STRICT_BYTES_OPTS) {
+    if (key in opts) {
+      const normalized = numBytesFromStr(String(opts[key]), true)
+      if (normalized === undefined) {
+        // Limitation: `opts` needs to have *some* value for this config key.
+        // However, it has already merged all sources of config, so we no longer
+        // have appropriate information to fallback to a lower-precedence valid
+        // value -- except for DEFAULTS. The *right* handling would
+        // validate/normalize values *before* merging.
+        logger.warn('invalid size value "%s" for "%s": falling back to default value %j',
+          opts[key], key, DEFAULTS[key])
+        opts[key] = numBytesFromStr(DEFAULTS[key], true)
+      } else {
+        opts[key] = normalized
+      }
+    }
   }
 }
 
@@ -566,14 +597,28 @@ function truncateOptions (opts) {
   if (opts.hostname) opts.hostname = truncate(String(opts.hostname), config.INTAKE_STRING_MAX_SIZE)
 }
 
-function bytes (input) {
+// Translate a string byte size, e.g. '10kb', into an integer number of bytes.
+//
+// If `strict` is `false` then eliding the units, e.g. '42', is allowed,
+// defaulting to 'b'. If the value is invalid, this returns `undefined`.
+function numBytesFromStr (input, strict) {
   const matches = input.match(/^(\d+)(b|kb|mb|gb)$/i)
-  if (!matches) return Number(input)
+  if (!matches) {
+    if (strict) {
+      return undefined
+    } else {
+      // Note: This currently allows float values and can return a NaN, which is
+      // unintentional.
+      // Not changing this for backwards compat. At the next major, the goal
+      // is to make all bytes-type config vars *strict*.
+      return Number(input)
+    }
+  }
 
   const suffix = matches[2].toLowerCase()
   let value = Number(matches[1])
 
-  if (!suffix || suffix === 'b') {
+  if (suffix === 'b') {
     return value
   }
 
@@ -687,6 +732,7 @@ function getBaseClientConfig (conf, agent) {
     // Sanitize conf
     truncateKeywordsAt: config.INTAKE_STRING_MAX_SIZE,
     truncateErrorMessagesAt: conf.errorMessageMaxLength,
+    truncateQueriesAt: conf.queryMaxLength,
 
     // HTTP conf
     secretToken: conf.secretToken,