[Filebeat] Fix RFC5424 date format in system/syslog in pipeline

CHANGELOG.next.asciidoc

@@ -119,6 +119,7 @@ https://github.com/elastic/beats/compare/v7.0.0-alpha2...master[Check the HEAD d
 - Require client_auth by default when ssl is enabled for tcp input {pull}12333[12333]
 - Require certificate authorities, certificate file, and key when SSL is enabled for the TCP input. {pull}12355[12355]
 - Load correct pipelines when system module is configured in modules.d. {pull}12340[12340]
+- Fix timezone offset parsing in system/syslog. {pull}12529[12529]
 
 *Heartbeat*
 

filebeat/module/system/syslog/ingest/pipeline.json

@@ -34,7 +34,10 @@
                 "formats": [
                     "MMM  d HH:mm:ss",
                     "MMM dd HH:mm:ss",
-                    "yyyy-MM-dd'T'HH:mm:ss.SSSSSSZZ"
+                    "yyyy-MM-dd'T'HH:mm:ss.SSSZZ",
+                    "yyyy-MM-dd'T'HH:mm:ss.SSSSSSZZ",
+                    "yyyy-MM-dd'T'HH:mm:ss.SSSXXX",
+                    "yyyy-MM-dd'T'HH:mm:ss.SSSSSSXXX"
                 ],
                 "ignore_failure": true
             }

filebeat/module/system/syslog/test/tz-offset.log

@@ -0,0 +1,2 @@
+1986-04-26T01:23:45.101+0400 rmbkmonitor04 shutdown[2649]: shutting down for system halt
+1986-04-26T01:23:45.388424+04:00 rmbkmonitor04 thermald: constraint_0_power_limit_uw exceeded.

filebeat/module/system/syslog/test/tz-offset.log-expected.json

@@ -0,0 +1,31 @@
+[
+    {
+        "@timestamp": "1986-04-25T21:23:45.101Z",
+        "event.dataset": "system.syslog",
+        "event.module": "system",
+        "event.timezone": "+00:00",
+        "fileset.name": "syslog",
+        "host.hostname": "rmbkmonitor04",
+        "input.type": "log",
+        "log.file.path": "tz-offset.log",
+        "log.offset": 0,
+        "message": "shutting down for system halt",
+        "process.name": "shutdown",
+        "process.pid": 2649,
+        "service.type": "system"
+    },
+    {
+        "@timestamp": "1986-04-25T21:23:45.388Z",
+        "event.dataset": "system.syslog",
+        "event.module": "system",
+        "event.timezone": "+00:00",
+        "fileset.name": "syslog",
+        "host.hostname": "rmbkmonitor04",
+        "input.type": "log",
+        "log.file.path": "tz-offset.log",
+        "log.offset": 89,
+        "message": "constraint_0_power_limit_uw exceeded.",
+        "process.name": "thermald",
+        "service.type": "system"
+    }
+]

filebeat/tests/system/test_modules.py

@@ -220,13 +220,32 @@ def clean_keys(obj):
     # ECS versions change for any ECS release, large or small
     ecs_key = ["ecs.version"]
 
+    # Keep source log filename for exceptions
+    filename = None
+    if "log.file.path" in obj:
+        filename = os.path.basename(obj["log.file.path"]).lower()
+
     for key in host_keys + time_keys + other_keys + ecs_key:
         delete_key(obj, key)
 
     # Remove timestamp for comparison where timestamp is not part of the log line
-    if (obj["event.dataset"] in ["icinga.startup", "redis.log", "haproxy.log", "system.auth", "system.syslog"]):
+    if (obj["event.dataset"] in ["icinga.startup", "redis.log", "haproxy.log", "system.auth"]):
         delete_key(obj, "@timestamp")
 
+    # HACK: This keeps @timestamp for the tz-offset.log in system.syslog.
+    #
+    # This can't be done for all syslog logs because most of them lack the year
+    # in their timestamp, so Elasticsearch will set it to the current year and
+    # that will cause the tests to fail every new year.
+    #
+    # The log.file.path key needs to be kept so that it is stored in the golden
+    # data, to prevent @timestamp to be removed from it before comparison.
+    if obj["event.dataset"] == "system.syslog":
+        if filename == "tz-offset.log":
+            obj["log.file.path"] = filename
+        else:
+            delete_key(obj, "@timestamp")
+
 
 def delete_key(obj, key):
     if key in obj: