public void when_serializing_logentries_can_serialize_valid_bulk_request_format()
{
// Note: converting an array does not create valid message for use in elasticsearch bulk operation
var actual = new ElasticsearchEventEntrySerializer("logstash", "slab", "instance", true).Serialize(new[] { CreateEventEntry(), CreateEventEntry() });
Assert.IsNotNull(actual);
Assert.IsTrue(this.IsValidBulkMessage(actual));
}
public void when_serializing_a_log_entry_then_object_can_serialize()
{
var payload = new Dictionary <string, object> {
{ "msg", "the message" }, { "date", DateTime.UtcNow }
};
var logObject =
EventEntryTestHelper.Create(
timestamp: DateTimeOffset.UtcNow,
payloadNames: payload.Keys,
payload: payload.Values);
var actual = new ElasticsearchEventEntrySerializer("logstash", "slab", "instance", true).Serialize(new[] { logObject });
Assert.IsNotNull(actual);
Assert.IsTrue(this.IsValidBulkMessage(actual));
}
public void when_serializing_a_log_entry_without_flattened_payload_then_payload_nested()
{
var payload = new Dictionary <string, object> {
{ "msg", "the message" }, { "date", DateTime.UtcNow }
};
var logObject =
EventEntryTestHelper.Create(
timestamp: DateTimeOffset.UtcNow,
payloadNames: payload.Keys,
payload: payload.Values);
var actual = new ElasticsearchEventEntrySerializer("logstash", "slab", "instance", false).Serialize(new[] { logObject });
var serializedEntry = actual.Split('\n')[1];
var jsonObject = JObject.Parse(serializedEntry);
Assert.IsTrue(jsonObject["Payload"]["msg"] != null);
Assert.IsTrue(jsonObject["Payload"]["date"] != null);
Assert.IsNotNull(actual);
Assert.IsTrue(this.IsValidBulkMessage(actual));
}
public void when_serializing_a_log_with_jsonprops_entry_then_json_merged()
{
var payload = new Dictionary <string, object>
{
{ "msg", "the message" },
{ "date", DateTime.UtcNow },
{ "_jsonPayload", "{\"test\": \"value\", \"test2\": \"value2\"}" }
};
var logObject =
EventEntryTestHelper.Create(
timestamp: DateTimeOffset.UtcNow,
payloadNames: payload.Keys,
payload: payload.Values);
var actual = new ElasticsearchEventEntrySerializer("logstash", "slab", "instance", true).Serialize(new[] { logObject });
// Make sure the payload is still valid
var obj = JObject.Parse(actual.Split('\n')[1]);
Assert.IsNotNull(actual);
Assert.IsTrue(this.IsValidBulkMessage(actual));
}
public void when_serializing_with_global_context_then_object_can_serialize()
{
var ctx = new Dictionary <string, string>()
{
{ "TestCtx1", "TestCtxValue1" },
{ "TestCtx2", "TestCtxValue2" }
};
var payload = new Dictionary <string, object> {
{ "msg", "the message" }, { "date", DateTime.UtcNow }
};
var logObject =
EventEntryTestHelper.Create(
timestamp: DateTimeOffset.UtcNow,
payloadNames: payload.Keys,
payload: payload.Values);
var actual = new ElasticsearchEventEntrySerializer("logstash", "slab", "instance", true, ctx).Serialize(new[] { logObject });
Debug.WriteLine(actual);
Assert.IsNotNull(actual);
Assert.IsTrue(this.IsValidBulkMessage(actual));
}
public void when_serializing_a_log_entry_with_activtyid_then_activityid_serialized()
{
var payload = new Dictionary <string, object> {
{ "msg", "the message" }, { "date", DateTime.UtcNow }
};
var logObject =
EventEntryTestHelper.Create(
timestamp: DateTimeOffset.UtcNow,
payloadNames: payload.Keys,
payload: payload.Values,
activityId: Guid.NewGuid(),
relatedActivityId: Guid.NewGuid());
var actual = new ElasticsearchEventEntrySerializer("logstash", "slab", "instance", true).Serialize(new[] { logObject });
var serializedEntry = actual.Split('\n')[1];
var jsonObject = JObject.Parse(serializedEntry);
Assert.AreEqual(logObject.ActivityId.ToString(), jsonObject["ActivityId"]);
Assert.AreEqual(logObject.RelatedActivityId.ToString(), jsonObject["RelatedActivityId"]);
Assert.IsNotNull(actual);
Assert.IsTrue(this.IsValidBulkMessage(actual));
}
/// <summary>
/// Initializes a new instance of the <see cref="ElasticsearchSink"/> class with the specified connection string and table address.
/// </summary>
/// <param name="instanceName">The name of the instance originating the entries.</param>
/// <param name="connectionString">The connection string for the storage account.</param>
/// <param name="index">Index name prefix formatted as index-{0:yyyy.MM.DD}.</param>
/// <param name="type">Elasticsearch entry type.</param>
/// <param name="flattenPayload">Flatten the payload collection when serializing event entries</param>
/// <param name="bufferInterval">The buffering interval to wait for events to accumulate before sending them to Elasticsearch.</param>
/// <param name="bufferingCount">The buffering event entry count to wait before sending events to Elasticsearch </param>
/// <param name="maxBufferSize">The maximum number of entries that can be buffered while it's sending to Windows Azure Storage before the sink starts dropping entries.</param>
/// <param name="onCompletedTimeout">Defines a timeout interval for when flushing the entries after an <see cref="OnCompleted"/> call is received and before disposing the sink.
/// This means that if the timeout period elapses, some event entries will be dropped and not sent to the store. Normally, calling <see cref="IDisposable.Dispose"/> on
/// the <see cref="Microsoft.Diagnostics.Tracing.EventListener"/> will block until all the entries are flushed or the interval elapses.
/// If <see langword="null"/> is specified, then the call will block indefinitely until the flush operation finishes.</param>
/// <param name="jsonGlobalContextExtension">A json encoded key/value set of global environment parameters to be included in each log entry</param>
public ElasticsearchSink(string instanceName, string connectionString, string index, string type, bool?flattenPayload, TimeSpan bufferInterval,
int bufferingCount, int maxBufferSize, TimeSpan onCompletedTimeout, string jsonGlobalContextExtension = null)
{
Guard.ArgumentNotNullOrEmpty(instanceName, "instanceName");
Guard.ArgumentNotNullOrEmpty(connectionString, "connectionString");
Guard.ArgumentNotNullOrEmpty(index, "index");
Guard.ArgumentNotNullOrEmpty(type, "type");
Guard.ArgumentIsValidTimeout(onCompletedTimeout, "onCompletedTimeout");
Guard.ArgumentGreaterOrEqualThan(0, bufferingCount, "bufferingCount");
string converted = ElasticsearchEventEntrySerializer.GetIndexName(index, DateTime.UtcNow);
if (Regex.IsMatch(converted, "[\\\\/*?\",<>|\\sA-Z]"))
{
throw new ArgumentException(Resource.InvalidElasticsearchIndexNameError, "index");
}
this.onCompletedTimeout = onCompletedTimeout;
this.instanceName = instanceName;
this.flattenPayload = flattenPayload ?? true;
this.elasticsearchUrl = new Uri(new Uri(connectionString), BulkServiceOperationPath);
this.index = index;
this.type = type;
var sinkId = string.Format(CultureInfo.InvariantCulture, "ElasticsearchSink ({0})", instanceName);
bufferedPublisher = new BufferedEventPublisher <EventEntry>(
sinkId,
this.PublishEventsAsync,
bufferInterval,
bufferingCount,
maxBufferSize,
this.cancellationTokenSource.Token);
this._jsonGlobalContextExtension = !string.IsNullOrEmpty(jsonGlobalContextExtension)? JsonConvert.DeserializeObject <Dictionary <string, string> >(jsonGlobalContextExtension): null;
}
public void when_serializing_a_log_entry_then_object_can_serialize_process_and_thread_id()
{
var payload = new Dictionary <string, object> {
{ "msg", "the message" }, { "date", DateTime.UtcNow }
};
var logObject =
EventEntryTestHelper.Create(
timestamp: DateTimeOffset.UtcNow,
payloadNames: payload.Keys,
payload: payload.Values,
processId: 300,
threadId: 500);
var actual = new ElasticsearchEventEntrySerializer("logstash", "slab", "instance", true).Serialize(new[] { logObject });
Assert.IsNotNull(actual);
Assert.IsTrue(this.IsValidBulkMessage(actual));
var serializedEntry = actual.Split('\n')[1];
var jsonObject = JObject.Parse(serializedEntry);
Assert.AreEqual(300, jsonObject["ProcessId"]);
Assert.AreEqual(500, jsonObject["ThreadId"]);
}
internal async Task <int> PublishEventsAsync(IList <EventEntry> collection)
{
try
{
string logMessages;
using (var serializer = new ElasticsearchEventEntrySerializer(this.index, this.type, this.instanceName, this.flattenPayload, this._jsonGlobalContextExtension))
{
logMessages = serializer.Serialize(collection);
}
var content = new StringContent(logMessages);
content.Headers.ContentType = new MediaTypeHeaderValue("application/json");
var response = await client.PostAsync(this.elasticsearchUrl, content, cancellationTokenSource.Token).ConfigureAwait(false);
// If there is an exception
if (response.StatusCode != HttpStatusCode.OK)
{
// Check the response for 400 bad request
if (response.StatusCode == HttpStatusCode.BadRequest)
{
var messagesDiscarded = collection.Count();
var errorContent = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
string serverErrorMessage;
// Try to parse the exception message
try
{
var errorObject = JObject.Parse(errorContent);
serverErrorMessage = errorObject["error"].Value <string>();
}
catch (Exception)
{
// If for some reason we cannot extract the server error message log the entire response
serverErrorMessage = errorContent;
}
// We are unable to write the batch of event entries - Possible poison message
// I don't like discarding events but we cannot let a single malformed event prevent others from being written
// We might want to consider falling back to writing entries individually here
SemanticLoggingEventSource.Log.CustomSinkUnhandledFault(string.Format("Elasticsearch sink unhandled exception {0} messages discarded with server error message {1}", messagesDiscarded, serverErrorMessage));
return(messagesDiscarded);
}
// This will leave the messages in the buffer
return(0);
}
var responseString = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
var responseObject = JObject.Parse(responseString);
var items = responseObject["items"] as JArray;
// If the response return items collection
if (items != null)
{
// NOTE: This only works with Elasticsearch 1.0
// Alternatively we could query ES as part of initialization check results or fall back to trying <1.0 parsing
// We should also consider logging errors for individual entries
//return items.Count(t => t["create"]["status"].Value<int>().Equals(201));
int count = items.Count();
return(count);
// Pre-1.0 Elasticsearch
// return items.Count(t => t["create"]["ok"].Value<bool>().Equals(true));
}
return(0);
}
catch (OperationCanceledException)
{
return(0);
}
catch (Exception ex)
{
// Although this is generally considered an anti-pattern this is not logged upstream and we have context
SemanticLoggingEventSource.Log.CustomSinkUnhandledFault(ex.ToString());
throw;
}
}
internal async Task <int> PublishEventsAsync(IList <EventEntry> collection)
{
try
{
string logMessages;
using (var serializer = new ElasticsearchEventEntrySerializer(this.index, this.type, this.instanceName, this.flattenPayload, this._jsonGlobalContextExtension))
{
logMessages = serializer.Serialize(collection);
}
var content = new StringContent(logMessages);
content.Headers.ContentType = new MediaTypeHeaderValue("application/json");
// buiding the basic authorization
if (this.userName != null && this.password != null)
{
string temp = this.userName + ":" + this.password;
var byteArray = Encoding.ASCII.GetBytes(temp);
var header = new AuthenticationHeaderValue("Basic", Convert.ToBase64String(byteArray));
this.client.DefaultRequestHeaders.Authorization = header;
}
var response = await this.client.PostAsync(this.elasticsearchUrl, content, this.cancellationTokenSource.Token).ConfigureAwait(false);
// If there is an exception
if (response.StatusCode != HttpStatusCode.OK)
{
// Check the response for 400 bad request
if (response.StatusCode == HttpStatusCode.BadRequest)
{
var messagesDiscarded = collection.Count();
var errorContent = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
string serverErrorMessage;
// Try to parse the exception message
try
{
var errorObject = JObject.Parse(errorContent);
serverErrorMessage = errorObject["error"].Value <string>();
}
catch (Exception)
{
// If for some reason we cannot extract the server error message log the entire response
serverErrorMessage = errorContent;
}
// We are unable to write the batch of event entries - Possible poison message
// I don't like discarding events but we cannot let a single malformed event prevent others from being written
// We might want to consider falling back to writing entries individually here
SemanticLoggingEventSource.Log.CustomSinkUnhandledFault(string.Format("Elasticsearch sink unhandled exception {0} messages discarded with server error message {1}", messagesDiscarded, serverErrorMessage));
return(messagesDiscarded);
}
// This will leave the messages in the buffer
return(0);
}
var responseString = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
var responseObject = JObject.Parse(responseString);
var items = responseObject["items"] as JArray;
// If the response return items collection
if (items != null)
{
// NOTE: This only works with Elasticsearch 1.0
// Alternatively we could query ES as part of initialization check results or fall back to trying <1.0 parsing
// We should also consider logging errors for individual entries
if (this.elasticsearchUrl.Port == 80 || this.elasticsearchUrl.Port == 443)
{
return(items.Count(t => t["index"]["status"].Value <int>().Equals(201)));
}
else if (this.elasticsearchUrl.Port == 9200 || this.elasticsearchUrl.Port == 9243)
{
//return items.Count(t => t["create"]["status"].Value<int>().Equals(201));
// Temporary fix: ES 2.3.1 returns create for bulk index, ES 5.0.2 and ES 5.1.1 return index for buld index.
// TODO Introduce version specific constant here.
return(items.Count(t => t["index"]["status"].Value <int>().Equals(201)));
}
// Pre-1.0 Elasticsearch
// return items.Count(t => t["create"]["ok"].Value<bool>().Equals(true));
}
return(0);
}
catch (OperationCanceledException)
{
return(0);
}
catch (Exception ex)
{
// Although this is generally considered an anti-pattern this is not logged upstream and we have context
Debug.WriteLine("{0} \n {1} \n {2} \n ", ex.Message, ex.InnerException, ex.StackTrace);
SemanticLoggingEventSource.Log.CustomSinkUnhandledFault(ex.ToString());
throw;
}
}
