Support a Sensu check custom attribute of alias, so that Sensu chec… (

#24) * Support a Sensu check custom attribute of `alias`, so that Sensu checks can specify the OpsGenie event ID value to use. This is necessary for supporting de-duplication of multiple Sensu alerts, based on a shared/common OpsGenie event ID. * Move the `alias` attribute into an `opsgenie` section. This helps indicate that this functionality is specific to this OpsGenie handler. It is also in line with the supported -j JSON config command-line option, which merges into a "check.opsgenie" namespace/object. * Nits for rubyists, per review feedback.
sensu-plugins · Jan 14, 2018 · 12170f5 · 12170f5
1 parent 4714256
commit 12170f5
Show file tree

Hide file tree

Showing 6 changed files with 79 additions and 7 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -4,6 +4,8 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 This CHANGELOG follows the format listed at [Keep A Changelog](http://keepachangelog.com/)
 
 ## [Unreleased]
+### Added
+- Added possibility to specify custom alias for event ID (@Castaglia)
 
 ## [4.0.1] - 2018-01-06
 ### Fixed

diff --git a/README.md b/README.md
@@ -79,3 +79,56 @@ To get this to work you need to specify a few different things. For a list of fi
 ```
 
 ## Notes
+
+If the check definition uses the custom `alias` attribute, _e.g._:
+```
+{
+  "checks": {
+    "check_mysql_access": {
+      "opsgenie": {
+        "alias": "MyCustomAlias",
+
+```
+then the `handler-opsgenie.rb` handler will use that attribute value as the
+OpsGenie event ID.  This can be useful for alert deduplication; checks on
+different clients for the same downstream resource can specify the same
+`alias` attribute, so that multiple alerts for the same resource are
+de-duplicated.
+
+By default, `handler-opsgenie.rb` creates an event ID from the client name
+and the check name.  Thus:
+```
+{
+  "checks": {
+    "check_mysql_access": {
+      "command": "/opt/sensu/embedded/bin/check-database.rb -h mysqldb",
+      "interval": 60,
+      "handlers": [ "opsgenie" ],
+      "standalone": true
+    }
+  }
+```
+running on a client named `web01` will create an alert using an event ID of
+`web01:check_mysql_access`.  And on a client named `web02`, it would create an
+alert with a _different_ event ID of `web02:check_mysql_access`, even though
+the `mysqldb` server being checked is the same for these clients.
+
+We can define a custom `alias` attribute in this check:
+```
+{
+  "checks": {
+    "check_mysql_access": {
+      "command": "/opt/sensu/embedded/bin/check-database.rb -h mysqldb",
+      "interval": 60,
+      "handlers": [ "opsgenie" ],
+      "standalone": true,
+
+      "opsgenie": {
+        "alias": "mysqldb"
+      }
+    }
+  }
+```
+And with this, running on multiple clients, any alerts would be generated
+with the same event ID of `mysqldb`, by using that `alias` attribute as the
+event ID.
diff --git a/bin/handler-opsgenie.rb b/bin/handler-opsgenie.rb
@@ -52,18 +52,18 @@ def process
                  end
       case response.code.to_s
       when '200', '202'
-        puts 'opsgenie -- ' + @event['action'].capitalize + 'd incident -- ' + event_id
+        puts "opsgenie -- #{@event['action'].capitalize}d incident -- #{event_id}"
       when '401'
-        puts 'opsgenie -- failed to ' + @event['action'] + ' incident -- not authorized'
+        puts "opsgenie -- failed to #{@event['action']} incident -- #{event_id}: not authorized"
       when '404'
-        puts 'opsgenie -- failed to ' + @event['action'] + ' incident -- ' + event_id + ' not found'
+        puts "opsgenie -- failed to #{@event['action']} incident -- #{event_id} not found"
       else
-        puts 'opsgenie -- failed to ' + @event['action'] + ' incident -- ' + event_id
+        puts "opsgenie -- failed to #{@event['action']} incident -- #{event_id}"
         puts "HTTP #{response.code} #{response.message}: #{response.body}"
       end
     end
   rescue Timeout::Error
-    puts 'opsgenie -- timed out while attempting to ' + @event['action'] + ' a incident -- ' + event_id
+    puts "opsgenie -- timed out while attempting to #{@event['action']} a incident -- #{event_id}"
   end
 
   def message
@@ -83,6 +83,8 @@ def default_message
   end
 
   def event_id
+    return @event['check']['opsgenie']['alias'] unless @event['check']['opsgenie'].nil? || @event['check']['opsgenie']['alias'].nil?
+
     # Do not use slashes in the event ID, as this alias becomes part of the URI
     # in the RESTful interactions with OpsGenie; use characters which can be
     # easily embedded into a URI.

diff --git a/test/fixtures/create-alert-with-alias.json b/test/fixtures/create-alert-with-alias.json
@@ -0,0 +1 @@
+{"client":{"name":"test01","address":"127.0.0.1","subscriptions":["all"],"timestamp":1326390159},"check":{"name":"some.fake.check.name","issued":1326390169,"output":"CRITICAL: text\n","status":2,"notification":"check failed","command":"/path/to/some/stuff/here -A do_smoke","subscribers":["all"],"interval":60,"handlers":["default","opsgenie"],"history":["0","0","2"],"flapping":false,"opsgenie":{"alias":"MY_CUSTOM_ALIAS"}},"occurrences":1,"action":"create"}
diff --git a/test/fixtures/resolve-alert-with-alias.json b/test/fixtures/resolve-alert-with-alias.json
@@ -0,0 +1 @@
+{"client":{"name":"test01","address":"127.0.0.1","subscriptions":["all"],"timestamp":1326390159},"check":{"name":"some.fake.check.name","issued":1326390169,"output":"CRITICAL: text\n","status":2,"notification":"check failed","command":"/path/to/some/stuff/here -A do_smoke","subscribers":["all"],"interval":60,"handlers":["default","opsgenie"],"history":["0","0","2"],"flapping":false,"opsgenie":{"alias":"MY_CUSTOM_ALIAS"}},"occurrences":1,"action":"resolve"}
diff --git a/test/integration/helpers/serverspec/handler-shared_spec.rb b/test/integration/helpers/serverspec/handler-shared_spec.rb
@@ -9,16 +9,29 @@
 create_alert_file = '/tmp/kitchen/data/test/fixtures/create-alert.json'
 resolve_alert_file = '/tmp/kitchen/data/test/fixtures/resolve-alert.json'
 
+create_alert_with_alias_file = '/tmp/kitchen/data/test/fixtures/create-alert-with-alias.json'
+resolve_alert_with_alias_file = '/tmp/kitchen/data/test/fixtures/resolve-alert-with-alias.json'
+
 # These tests would require a valid OpsGenie API key and heartbeat name
 # configured in order to succeed.  Thus for now, we limit ourselves to the
 # expected failure cases.
 
 describe 'handler-opsgenie' do
+  default_event_id_pattern = 'test01:some\.fake\.check\.name'
   describe command("#{handler} < #{create_alert_file}") do
-    its(:stdout) { should match(/failed to create incident.*not authorized/) }
+    its(:stdout) { should match(/failed to create incident.*#{default_event_id_pattern}.*not authorized/) }
   end
 
   describe command("#{handler} < #{resolve_alert_file}") do
-    its(:stdout) { should match(/failed to resolve incident.*not authorized/) }
+    its(:stdout) { should match(/failed to resolve incident.*#{default_event_id_pattern}.*not authorized/) }
+  end
+
+  custom_event_id_pattern = 'MY_CUSTOM_ALIAS'
+  describe command("#{handler} < #{create_alert_with_alias_file}") do
+    its(:stdout) { should match(/failed to create incident.*#{custom_event_id_pattern}.*not authorized/) }
+  end
+
+  describe command("#{handler} < #{resolve_alert_with_alias_file}") do
+    its(:stdout) { should match(/failed to resolve incident.*#{custom_event_id_pattern}.*not authorized/) }
   end
 end