qlog: Structured Logging for Network Protocols

1.1. Conventions and Terminology

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶

Serialization examples in this document use JSON ([JSON]) unless otherwise indicated.¶

1.2. Use of CDDL

To define events and data structures, all qlog documents use the Concise Data Definition Language [CDDL]. This document uses the basic syntax, the specific text, uint, float32, float64, bool, and any types, as well as the .default, .size, and .regexp control operators, the ~ unwrapping operator, and the $ and $$ extension points syntax from [CDDL].¶

Additionally, this document defines the following custom types for clarity:¶

; CDDL's uint is defined as being 64-bit in size
; but for many protocol fields it is better to be restrictive
; and explicit
uint8 = uint .size 1
uint16 = uint .size 2
uint32 = uint .size 4
uint64 = uint .size 8

; an even-length lowercase string of hexadecimally encoded bytes
; examples: 82dc, 027339, 4cdbfd9bf0
; this is needed because the default CDDL binary string (bytes/bstr)
; is only CBOR and not JSON compatible
hexstring = text .regexp "([0-9a-f]{2})*"

All timestamps and time-related values (e.g., offsets) in qlog are logged as float64 in the millisecond resolution.¶

Other qlog documents can define their own CDDL-compatible (struct) types (e.g., separately for each Packet type that a protocol supports).¶

The ordering of member fields in qlog CDDL type definitions is not significant. The ordering of member fields in the serialization formats defined in this document, JSON (Section 11.1) and JSON Text Sequences (Section 11.2), is not significant and qlog tools MUST NOT assume so. Other qlog serialization formats MAY define field order significance, if they do they MUST define requirements for qlog tools supporting those formats.¶

• Note to RFC editor: Please remove the following text in this section before publication.¶

The main general CDDL syntax conventions in this document a reader should be aware of for easy reading comprehension are:¶

• ? obj : this object is optional¶
• TypeName1 / TypeName2 : a union of these two types (object can be either type 1 OR type 2)¶
• obj: TypeName : this object has this concrete type¶
• obj: [* TypeName] : this object is an array of this type with minimum size of 0 elements¶
• obj: [+ TypeName] : this object is an array of this type with minimum size of 1 element¶
• TypeName = ... : defines a new type¶
• EnumName = "entry1" / "entry2" / entry3 / ...: defines an enum¶
• StructName = { ... } : defines a new struct type¶
• ; : single-line comment¶
• * text => any : special syntax to indicate 0 or more fields that have a string key that maps to any value. Used to indicate a generic JSON object.¶

All timestamps and time-related values (e.g., offsets) in qlog are logged as float64 in the millisecond resolution.¶

Other qlog documents can define their own CDDL-compatible (struct) types (e.g., separately for each Packet type that a protocol supports).¶

3.1. Concrete Log File Schema URIs

Concrete log file schemas MUST identify themselves using a URI [RFC3986].¶

Log file schemas defined by RFCs MUST register a URI in the "qlog log file schema URIs" registry and SHOULD use a URN of the form urn:ietf:params:qlog:file:<schema-identifier>, where <schema-identifier> is a globally-unique text name using only characters in the URI unreserved range; see Section 2.3 of [RFC3986]. This document registers urn:ietf:params:qlog:file:contained (Section 4) and urn:ietf:params:qlog:file:sequential (Section 5).¶

Private or non-standard log file schemas MAY register a URI in the "qlog log file schema URIs" registry but MUST NOT use a URN of the form urn:ietf:params:qlog:file:<schema-identifier>. URIs that contain a domain name SHOULD also contain a month-date in the form mmyyyy. For example, "https://example.org/072024/globallyuniquelogfileschema". The definition of the log file schema and assignment of the URI MUST have been authorized by the owner of the domain name on or very close to that date. This avoids problems when domain names change ownership. The URI does not need to be dereferencable, allowing for confidential use or to cover the case where the log file schema continues to be used after the organization that defined them ceases to exist.¶

The "qlog log file schema URIs" registry operates under the Expert Review policy, per Section 4.5 of [RFC8126]. When reviewing requests, the expert MUST check that the URI is appropriate to the concrete log file schema and satisfies the requirements in this section. A request to register a private or non-standard log file schema URI using a URN of the form urn:ietf:params:qlog:file:<schema-identifier> MUST be rejected.¶

Registration requests should use the following template:¶

Log File Schema URI:

[the log file schema identifier]¶

Description:

[a description of the log file schema]¶

Reference:

[to a specification defining the log file schema]¶

4.1. Traces

It can be advantageous to group several related qlog traces together in a single file. For example, it is possible to simultaneously perform logging on the client, on the server, and on a single point on their common network path. For analysis, it is useful to aggregate these three individual traces together into a single file, so it can be uniquely stored, transferred, and annotated.¶

The QlogFile "traces" field is an array that contains a list of individual qlog traces. When capturing a qlog at a vantage point, it is expected that the traces field contains a single entry. Files can be aggregated, for example as part of a post-processing operation, by copying the traces in component to files into the combined "traces" array of a new, aggregated qlog file.¶

4.2. Trace

The exact conceptual definition of a Trace can be fluid. For example, a trace could contain all events for a single connection, for a single endpoint, for a single measurement interval, for a single protocol, etc. In the normal use case however, a trace is a log of a single data flow collected at a single location or vantage point. For example, for QUIC, a single trace only contains events for a single logical QUIC connection for either the client or the server.¶

A Trace contains some metadata in addition to qlog events, defined in Figure 5 as:¶

Trace = {
    ? title: text
    ? description: text
    ? common_fields: CommonFields
    ? vantage_point: VantagePoint
    events: [* Event]
}

The optional "title" and "description" fields provide additional free-text information about the trace.¶

The optional "common_fields" field is described in Section 7.7.¶

The optional "vantage_point" field is described in Section 6.¶

The semantics and context of the trace can mainly be deduced from the entries in the "common_fields" list and "vantage_point" field.¶

JSON serialization example:¶

{
    "title": "Name of this particular trace (short)",
    "description": "Description for this trace (long)",
    "common_fields": {
        "ODCID": "abcde1234",
        "time_format": "absolute"
    },
    "vantage_point": {
        "name": "backend-67",
        "type": "server"
    },
    "events": [...]
}

4.3. TraceError

A TraceError indicates that an attempt to find/convert a file for inclusion in the aggregated qlog was made, but there was an error during the process. Rather than silently dropping the erroneous file, it can be explicitly included in the qlog file as an entry in the "traces" array, defined in Figure 7 as:¶

TraceError = {
    error_description: text

    ; the original URI used for attempted find of the file
    ? uri: text
    ? vantage_point: VantagePoint
}

JSON serialization example:¶

{
    "error_description": "File could not be found",
    "uri": "/srv/traces/today/latest.qlog",
    "vantage_point": { type: "server" }
}

Note that another way to combine events of different traces in a single qlog file is through the use of the "group_id" field, discussed in Section 7.5.¶

5.1. TraceSeq

TraceSeq is used with QlogFileSeq. It is conceptually similar to a Trace, with the exception that qlog events are not contained within it, but rather appended after it in a QlogFileSeq.¶

TraceSeq = {
    ? title: text
    ? description: text
    ? common_fields: CommonFields
    ? vantage_point: VantagePoint
}

7.1. Timestamps

An event's "time" field indicates the timestamp at which the event occurred. Its value is typically the Unix timestamp since the 1970 epoch (number of milliseconds since midnight UTC, January 1, 1970, ignoring leap seconds). However, qlog supports two more succinct timestamps formats to allow reducing file size. The employed format is indicated in the "time_format" field, which allows one of three values: "absolute", "delta" or "relative".¶

TimeFormat = "absolute" /
             "delta" /
             "relative"

• Absolute: Include the full absolute timestamp with each event. This approach uses the largest amount of characters. This is also the default value of the "time_format" field.¶
• Delta: Delta-encode each time value on the previously logged value. The first event in a trace typically logs the full absolute timestamp. This approach uses the least amount of characters.¶
• Relative: Specify a full "reference_time" timestamp (typically this is done up-front in "common_fields", see Section 7.7) and include only relatively-encoded values based on this reference_time with each event. The "reference_time" value is typically the first absolute timestamp. This approach uses a medium amount of characters.¶

The first option is good for stateless loggers, the second and third for stateful loggers. The third option is generally preferred, since it produces smaller files while being easier to reason about. An example for each option can be seen in Figure 17.¶

The absolute approach will use:
1500, 1505, 1522, 1588

The delta approach will use:
1500, 5, 17, 66

The relative approach will:
- set the reference_time to 1500 in "common_fields"
- use: 0, 5, 22, 88

One of these options is typically chosen for the entire trace (put differently: each event has the same value for the "time_format" field). Each event MUST include a timestamp in the "time" field.¶

Events in each individual trace SHOULD be logged in strictly ascending timestamp order (though not necessarily absolute value, for the "delta" format). Tools MAY sort all events on the timestamp before processing them, though are not required to (as this could impose a significant processing overhead). This can be a problem especially for multi-threaded and/or streaming loggers, who could consider using a separate post-processor to order qlog events in time if a tool do not provide this feature.¶

Timestamps do not have to use the UNIX epoch timestamp as their reference. For example for privacy considerations, any initial reference timestamps (for example "endpoint uptime in ms" or "time since connection start in ms") can be chosen. Tools SHOULD NOT assume the ability to derive the absolute Unix timestamp from qlog traces, nor allow on them to relatively order events across two or more separate traces (in this case, clock drift should also be taken into account).¶

7.2. Path

A qlog event can be associated with a single "network path" (usually, but not always, identified by a 4-tuple of IP addresses and ports). In many cases, the path will be the same for all events in a given trace, and does not need to be logged explicitly with each event. In this case, the "path" field can be omitted (in which case the default value of "" is assumed) or reflected in "common_fields" instead (see Section 7.7).¶

However, in some situations, such as during QUIC's Connection Migration or when using Multipath features, it is useful to be able to split events across multiple (concurrent) paths.¶

Definition:¶

PathID = text .default ""

The "path" field is an identifier that is associated with a single network path. This document intentionally does not define further how to choose this identifier's value per-path or how to potentially log other parameters that can be associated with such a path. This is left for other documents. Implementers are free to encode path information directly into the PathID or to log associated info in a separate event. For example, QUIC has the "path_assigned" event to couple the PathID value to a specific path configuration, see [QLOG-QUIC].¶

7.3. ProtocolTypeList and ProtocolType

An event's "protocol_type" array field indicates to which protocols (or protocol "stacks") this event belongs. This allows a single qlog file to aggregate traces of different protocols (e.g., a web server offering both TCP+HTTP/2 and QUIC+HTTP/3 connections).¶

ProtocolTypeList = [+ $ProtocolType]

$ProtocolType /= "UNKNOWN"

For example, QUIC and HTTP/3 events have the "QUIC" and "HTTP3" protocol_type entry values, see [QLOG-QUIC] and [QLOG-H3].¶

Typically however, all events in a single trace are of the same few protocols, and this array field is logged once in "common_fields", see Section 7.7.¶

7.4. Triggers

Sometimes, additional information is needed in the case where a single event can be caused by a variety of other events. In the normal case, the context of the surrounding log messages gives a hint as to which of these other events was the cause. However, in highly-parallel and optimized implementations, corresponding log messages might separated in time. Another option is to explicitly indicate these "triggers" in a high-level way per-event to get more fine-grained information without much additional overhead.¶

In qlog, the optional "trigger" field contains a string value describing the reason (if any) for this event instance occurring, see Section 8.2. While this "trigger" field could be a property of the qlog Event itself, it is instead a property of the "data" field instead. This choice was made because many event types do not include a trigger value, and having the field at the Event-level would cause overhead in some serializations. Additional information on the trigger can be added in the form of additional member fields of the "data" field value, yet this is highly implementation-specific, as are the trigger field's string values.¶

One purely illustrative example of some potential triggers for QUIC's "packet_dropped" event is shown in Figure 20:¶

TransportPacketDropped = {
    ? packet_type: PacketType
    ? raw_length: uint16
    ? trigger: "key_unavailable" /
               "unknown_connection_id" /
               "decrypt_error" /
               "unsupported_version"
}

7.5. Grouping

As discussed in Section 4.2, a single qlog file can contain several traces taken from different vantage points. However, a single trace from one endpoint can also contain events from a variety of sources. For example, a server implementation might choose to log events for all incoming connections in a single large (streamed) qlog file. As such, a method for splitting up events belonging to separate logical entities is required.¶

The simplest way to perform this splitting is by associating a "group id" to each event that indicates to which conceptual "group" each event belongs. A post-processing step can then extract events per group. However, this group identifier can be highly protocol and context-specific. In the example above, the QUIC "Original Destination Connection ID" could be used to uniquely identify a connection. As such, they might add a "ODCID" field to each event. However, a middlebox logging IP or TCP traffic might rather use four-tuples to identify connections, and add a "four_tuple" field.¶

As such, to provide consistency and ease of tooling in cross-protocol and cross-context setups, qlog instead defines the common "group_id" field, which contains a string value. Implementations are free to use their preferred string serialization for this field, so long as it contains a unique value per logical group. Some examples can be seen in Figure 22.¶

GroupID = text

JSON serialization example for events grouped by four tuples and QUIC connection IDs:¶

"events": [
    {
        "time": 1553986553579,
        "protocol_type": ["TCP", "TLS", "HTTP2"],
        "group_id": "ip1=2001:67c:1232:144:9498:6df6:f450:110b,
                   ip2=2001:67c:2b0:1c1::198,port1=59105,port2=80",
        "name": "quic:packet_received",
        "data": { ... }
    },
    {
        "time": 1553986553581,
        "protocol_type": ["QUIC","HTTP3"],
        "group_id": "127ecc830d98f9d54a42c4f0842aa87e181a",
        "name": "quic:packet_sent",
        "data": { ... }
    }
]

Note that in some contexts (for example a Multipath transport protocol) it might make sense to add additional contextual per-event fields (for example "path_id"), rather than use the group_id field for that purpose.¶

Note also that, typically, a single trace only contains events belonging to a single logical group (for example, an individual QUIC connection). As such, instead of logging the "group_id" field with an identical value for each event instance, this field is typically logged once in "common_fields", see Section 7.7.¶

7.6. SystemInformation

The "system_info" field can be used to record system-specific details related to an event. This is useful, for instance, where an application splits work across CPUs, processes, or threads and events for a single trace occur on potentially different combinations thereof. Each field is optional to support deployment diversity.¶

SystemInformation = {
  ? processor_id: uint32
  ? process_id: uint32
  ? thread_id: uint32
}

7.7. CommonFields

As discussed in the previous sections, information for a typical qlog event varies in three main fields: "time", "name" and associated data. Additionally, there are also several more advanced fields that allow mixing events from different protocols and contexts inside of the same trace (for example "protocol_type" and "group_id"). In most "normal" use cases however, the values of these advanced fields are consistent for each event instance (for example, a single trace contains events for a single QUIC connection).¶

To reduce file size and making logging easier, qlog uses the "common_fields" list to indicate those fields and their values that are shared by all events in this component trace. This prevents these fields from being logged for each individual event. An example of this is shown in Figure 23.¶

JSON serialization with repeated field values
per-event instance:

{
    "events": [{
            "group_id": "127ecc830d98f9d54a42c4f0842aa87e181a",
            "protocol_type": ["QUIC","HTTP3"],
            "time_format": "relative",
            "reference_time": 1553986553572,

            "time": 2,
            "name": "quic:packet_received",
            "data": { ... }
        },{
            "group_id": "127ecc830d98f9d54a42c4f0842aa87e181a",
            "protocol_type": ["QUIC","HTTP3"],
            "time_format": "relative",
            "reference_time": 1553986553572,

            "time": 7,
            "name": "http:frame_parsed",
            "data": { ... }
        }
    ]
}

JSON serialization with repeated field values instead
extracted to common_fields:

{
    "common_fields": {
        "group_id": "127ecc830d98f9d54a42c4f0842aa87e181a",
        "protocol_type": ["QUIC","HTTP3"],
        "time_format": "relative",
        "reference_time": 1553986553572
    },
    "events": [
        {
            "time": 2,
            "name": "quic:packet_received",
            "data": { ... }
        },{
            "time": 7,
            "name": "http:frame_parsed",
            "data": { ... }
        }
    ]
}

An event's "common_fields" field is a generic dictionary of key-value pairs, where the key is always a string and the value can be of any type, but is typically also a string or number. As such, unknown entries in this dictionary MUST be disregarded by the user and tools (i.e., the presence of an unknown field is explicitly NOT an error).¶

The list of default qlog fields that are typically logged in common_fields (as opposed to as individual fields per event instance) are shown in the listing below:¶

CommonFields = {
    ? path: PathID
    ? time_format: TimeFormat
    ? reference_time: float64
    ? protocol_type: ProtocolTypeList
    ? group_id: GroupID
    * text => any
}

Tools MUST be able to deal with these fields being defined either on each event individually or combined in common_fields. Note that if at least one event in a trace has a different value for a given field, this field MUST NOT be added to common_fields but instead defined on each event individually. Good example of such fields are "time" and "data", who are divergent by nature.¶

8.1. Event Schema and Event Category URIs

Event schema defined by RFCs MUST register all categories in the "qlog event category URIs" registry and SHOULD use a URN of the form urn:ietf:params:qlog:events:<namespace identifier>#<category identifier>, where <category identifier> is globally unique. For example, this document defines the generic event schema (Section 9) that uses the gen namespace containing the loglevel and sim categories. Other examples of event schema define the quic [QLOG-QUIC] and http [QLOG-H3] namespaces.¶

Private or non-standard event categories MAY be registered in the "qlog event category URIs" registry but MUST NOT use a URN of the form urn:ietf:params:qlog:events:<namespace identifier>#<category identifier>. URIs that contain a domain name SHOULD also contain a month-date in the form mmyyyy. For example, "https://example.org/072024/customeventschema#globallyuniquecategory". The definition of the category and assignment of the URI MUST have been authorized by the owner of the domain name on or very close to that date. This avoids problems when domain names change ownership. The URI does not need to be dereferencable, allowing for confidential use or to cover the case where the log file schema continues to be used after the organization that defined them ceases to exist.¶

The "qlog event category URIs" registry operates under the Expert Review policy, per Section 4.5 of [RFC8126]. When reviewing requests, the expert MUST check that the URI is appropriate to the event schema and satisfies the requirements in Section 8 and this section. A request to register a private or non-standard category URI using a URN of the form urn:ietf:params:qlog:events:<namespace identifier>#<category identifier MUST be rejected.¶

Registration requests should use the following template:¶

Event Category URI:

[the event category identifier]¶

Description:

[a description of the event category]¶

Reference:

[to a specification defining the event schema defining the event category]¶

8.2. Extending the Data Field

An event's "data" field is a generic key-value map (e.g., JSON object). It defines the per-event metadata that is to be logged. Its specific subfields and their semantics are defined per concrete event type. For example, data field definitions for QUIC and HTTP/3 can be found in [QLOG-QUIC] and [QLOG-H3].¶

In order to keep qlog fully extensible, two separate CDDL extension points ("sockets" or "plugs") are used to fully define data fields.¶

Firstly, to allow existing data field definitions to be extended (for example by adding an additional field needed for a new protocol feature), a CDDL "group socket" is used. This takes the form of a subfield with a name of * $$CATEGORY-NAME-extension. This field acts as a placeholder that can later be replaced with newly defined fields by assigning them to the socket with the //= operator. Multiple extensions can be assigned to the same group socket. An example is shown in Figure 26.¶

; original definition in document A
MyCategoryEventX = {
    field_a: uint8

    * $$mycategory-eventx-extension
}

; later extension of EventX in document B
$$mycategory-eventx-extension //= (
  ? additional_field_b: bool
)

; another extension of EventX in document C
$$mycategory-eventx-extension //= (
  ? additional_field_c: text
)

; if document A, B and C are then used in conjunction,
; the combined MyCategoryEventX CDDL is equivalent to this:
MyCategoryEventX = {
    field_a: uint8

    ? additional_field_b: bool
    ? additional_field_c: text
}

Secondly, to allow documents to define fully new event data field definitions (as opposed to extend existing ones), a CDDL "type socket" is used. For this purpose, the type of the "data" field in the qlog Event type (see Figure 14) is the extensible $ProtocolEventData type. This field acts as an open enum of possible types that are allowed for the data field. As such, any new event data field is defined as its own CDDL type and later merged with the existing $ProtocolEventData enum using the /= extension operator. Any generic key-value map type can be assigned to $ProtocolEventData (the only common "data" subfield defined in this document is the optional trigger field, see Section 7.4). An example of this setup is shown in Figure 27.¶

; We define two separate events in a single document
MyCategoryEvent1 /= {
    field_1: uint8

    ? trigger: text

    * $$mycategory-event1-extension
}

MyCategoryEvent2 /= {
    field_2: bool

    ? trigger: text

    * $$mycategory-event2-extension
}

; the events are both merged with the existing
; $ProtocolEventData type enum
$ProtocolEventData /= MyCategoryEvent1 / MyCategoryEvent2

; the "data" field of a qlog event can now also be of type
; MyCategoryEvent1 and MyCategoryEvent2

Documents defining new qlog events MUST properly extend $ProtocolEventData when defining data fields to enable automated validation of aggregated qlog schemas. Furthermore, they SHOULD properly add a * $$CATEGORY-NAME-extension extension field to newly defined event data to allow the new events to be properly extended by other documents.¶

A combined but purely illustrative example of the use of both extension points for a conceptual QUIC "packet_sent" event is shown in Figure 28:¶

; defined in the main QUIC event document
TransportPacketSent = {
    ? packet_size: uint16
    header: PacketHeader
    ? frames:[* QuicFrame]
    ? trigger: "pto_probe" /
               "retransmit_timeout" /
               "bandwidth_probe"

    * $$transport-packetsent-extension
}

; Add the event to the global list of recognized qlog events
$ProtocolEventData /= TransportPacketSent

; Defined in a separate document that describes a
; theoretical QUIC protocol extension
$$transport-packetsent-extension //= (
  ? additional_field: bool
)

; If both documents are utilized at the same time,
; the following JSON serialization would pass an automated
; CDDL schema validation check:

{
  "time": 123456,
  "category": "transport",
  "name": "packet_sent",
  "data": {
      "packet_size": 1280,
      "header": {
          "packet_type": "1RTT",
          "packet_number": 123
      },
      "frames": [
          {
              "frame_type": "stream",
              "length": 1000,
              "offset": 456
          },
          {
              "frame_type": "padding"
          }
      ],
      additional_field: true
  }
}

8.3. Event Importance Levels

Depending on how events are designed, it may be that several events allow the logging of similar or overlapping data. For example the separate QUIC connection_started event overlaps with the more generic connection_state_updated. In these cases, it is not always clear which event should be logged or used, and which event should take precedence if e.g., both are present and provide conflicting information.¶

To aid in this decision making, qlog defines three event importance levels, in decreasing order of importance and expected usage:¶

• Core¶
• Base¶
• Extra¶

Concrete event types SHOULD define an importance level.¶

Core-level events SHOULD be present in all qlog files for a given protocol. These are typically tied to basic packet and frame parsing and creation, as well as listing basic internal metrics. Tool implementers SHOULD expect and add support for these events, though SHOULD NOT expect all Core events to be present in each qlog trace.¶

Base-level events add additional debugging options and MAY be present in qlog files. Most of these can be implicitly inferred from data in Core events (if those contain all their properties), but for many it is better to log the events explicitly as well, making it clearer how the implementation behaves. These events are for example tied to passing data around in buffers, to how internal state machines change, and used to help show when decisions are actually made based on received data. Tool implementers SHOULD at least add support for showing the contents of these events, if they do not handle them explicitly.¶

Extra-level events are considered mostly useful for low-level debugging of the implementation, rather than the protocol. They allow more fine-grained tracking of internal behavior. As such, they MAY be present in qlog files and tool implementers MAY add support for these, but they are not required to.¶

Note that in some cases, implementers might not want to log for example data content details in Core-level events due to performance or privacy considerations. In this case, they SHOULD use (a subset of) relevant Base-level events instead to ensure usability of the qlog output. As an example, implementations that do not log QUIC packet_received events and thus also not which (if any) ACK frames the packet contains, SHOULD log packets_acked events instead.¶

Finally, for event types whose data (partially) overlap with other event types' definitions, where necessary the event definition document should include explicit guidance on which to use in specific situations.¶

8.4. Tooling Expectations

qlog is an extensible format and it is expected that new event schema will emerge that define new categories, event types, event fields (e.g., a field indicating an event's privacy properties), as well as values for the "trigger" property within the "data" field, or other member fields of the "data" field, as they see fit.¶

It SHOULD NOT be expected that general-purpose tools will recognize or visualize all forms of qlog extension. Tools SHOULD allow for the presence of unknown event fields and make an effort to visualize even unknown data if possible, otherwise they MUST ignore it.¶

8.5. Further Design Guidance

There are several ways of defining concrete event types. In practice, two main types of approach have been observed: a) those that map directly to concepts seen in the protocols (e.g., packet_sent) and b) those that act as aggregating events that combine data from several possible protocol behaviors or code paths into one (e.g., parameters_set). The latter are typically used as a means to reduce the amount of unique event definitions, as reflecting each possible protocol event as a separate qlog entity would cause an explosion of event types.¶

Additionally, logging duplicate data is typically prevented as much as possible. For example, packet header values that remain consistent across many packets are split into separate events (for example spin_bit_updated or connection_id_updated for QUIC).¶

Finally, when logging additional state change events, those state changes can often be directly inferred from data on the wire (for example flow control limit changes). As such, if the implementation is bug-free and spec-compliant, logging additional events is typically avoided. Exceptions have been made for common events that benefit from being easily identifiable or individually logged (for example packets_acked).¶

9.1. Log Level Events

In typical logging setups, users utilize a discrete number of well-defined logging categories, levels or severities to log freeform (string) data. The log level event category replicates this approach to allow implementations to fully replace their existing text-based logging by qlog. This is done by providing events to log generic strings for the typical well-known logging levels (error, warning, info, debug, verbose). The category identifier is "loglevel". The category URI is is urn:ietf:params:qlog:events:gen#loglevel.¶

LogLevelEventData = LogLevelError /
                    LogLevelWarning /
                    LogLevelInfo /
                    LogLevelDebug /
                    LogLevelVerbose

$ProtocolEventData /= LogLevelEventData

The event types are further defined below, their identifier is the heading name.¶

LogLevelError = {
    ? code: uint64
    ? message: text

    * $$loglevel-error-extension
}

LogLevelWarning = {
    ? code: uint64
    ? message: text

    * $$loglevel-warning-extension
}

LogLevelInfo = {
    message: text

    * $$loglevel-info-extension
}

LogLevelDebug = {
    message: text

    * $$loglevel-debug-extension
}

LogLevelVerbose = {
    message: text

    * $$loglevel-verbose-extension
}

9.2. Simulation Events

When evaluating a protocol implementation, one typically sets up a series of interoperability or benchmarking tests, in which the test situations can change over time. For example, the network bandwidth or latency can vary during the test, or the network can be fully disable for a short time. In these setups, it is useful to know when exactly these conditions are triggered, to allow for proper correlation with other events. This category defines event types to allow logging of such simulation metadata and its identifier is "sim". The category URI is urn:ietf:params:qlog:events:gen#sim.¶

SimulationEventData = SimulationScenario /
                SimulationMarker

$ProtocolEventData /= SimulationEventData

The event types are further defined below, their identifier is the heading name.¶

SimulationScenario = {
    ? name: text
    ? details: {* text => any }

    * $$simulation-scenario-extension
}

SimulationMarker = {
    ? type: text
    ? message: text

    * $$simulation-marker-extension
}

11.1. qlog to JSON mapping

As described in Section 4, JSON is the default qlog serialization. When mapping qlog to normal JSON, QlogFile (Figure 3) is used. The Media Type is "application/qlog+json" per [RFC6839]. The file extension/suffix SHOULD be ".qlog".¶

In accordance with Section 8.1 of [RFC8259], JSON files are required to use UTF-8 both for the file itself and the string values it contains. In addition, all qlog field names MUST be lowercase when serialized to JSON.¶

In order to serialize CDDL-based qlog event and data structure definitions to JSON, the official CDDL-to-JSON mapping defined in Appendix E of [CDDL] SHOULD be employed.¶

11.2. qlog to JSON Text Sequences mapping

One of the downsides of using normal JSON is that it is inherently a non-streamable format. A qlog serializer could work around this by opening a file, writing the required opening data, streaming qlog events by appending them, and then finalizing the log by appending appropriate closing tags e.g., "]}]}". However, failure to append closing tags, could lead to problems because most JSON parsers will fail if a document is malformed. Some streaming JSON parsers are able to handle missing closing tags, however they are not widely deployed in popular environments (e.g., Web browsers)¶

To overcome the issues related to JSON streaming, a qlog mapping to a streamable JSON format called JSON Text Sequences (JSON-SEQ) ([RFC7464]) is provided.¶

JSON Text Sequences are very similar to JSON, except that objects are serialized as individual records, each prefixed by an ASCII Record Separator (<RS>, 0x1E), and each ending with an ASCII Line Feed character (\n, 0x0A). Note that each record can also contain any amount of newlines in its body, as long as it ends with a newline character before the next <RS> character.¶

In order to leverage the streaming capability, each qlog event is serialized and interpreted as an individual JSON Text Sequence record, that is appended as a new object to the back of an event stream or log file. Put differently, unlike default JSON, it does not require a document to be wrapped as a full object with "{ ... }" or "[... ]".¶

This alternative record streaming approach cannot be accommodated by QlogFile (Figure 3). Instead, QlogFileSeq is defined in Figure 9, which notably includes only a single trace (TraceSeq) and omits an explicit "events" array. An example is provided in Figure 10. The "group_id" field can still be used on a per-event basis to include events from conceptually different sources in a single JSON-SEQ qlog file.¶

When mapping qlog to JSON-SEQ, the Media Type is "application/qlog+json-seq" per [RFC8091]. The file extension/suffix SHOULD be ".sqlog" (for "streaming" qlog).¶

While not specifically required by the JSON-SEQ specification, all qlog field names MUST be lowercase when serialized to JSON-SEQ.¶

In order to serialize all other CDDL-based qlog event and data structure definitions to JSON-SEQ, the official CDDL-to-JSON mapping defined in Appendix E of [CDDL] SHOULD be employed.¶

11.3. JSON Interoperability

Some JSON implementations have issues with the full JSON format, especially those integrated within a JavaScript environment (e.g., Web browsers, NodeJS). I-JSON (Internet-JSON) is a subset of JSON for such environments; see [I-JSON]. One of the key limitations of JavaScript, and thus I-JSON, is that it cannot represent full 64-bit integers in standard operating mode (i.e., without using BigInt extensions), instead being limited to the range -(2⁵³)+1 to (2⁵³)-1.¶

To accommodate such constraints in CDDL, Appendix E of [CDDL] recommends defining new CDDL types for int64 and uint64 that limit their values to the restricted 64-bit integer range. However, some of the protocols that qlog is intended to support (e.g., QUIC, HTTP/3), can use the full range of uint64 values.¶

As such, to support situations where I-JSON is in use, seralizers MAY encode uint64 values using JSON strings. qlog parsers, therefore, SHOULD support parsing of uint64 values from JSON strings or JSON numbers unless there is out-of-band information indicating that neither the serializer nor parser are constrained by I-JSON.¶

11.4. Truncated values

For some use cases (e.g., limiting file size, privacy), it can be necessary not to log a full raw blob (using the hexstring type) but instead a truncated value. For example, one might only store the first 100 bytes of an HTTP response body to be able to discern which file it actually contained. In these cases, the original byte-size length cannot be obtained from the serialized value directly.¶

As such, all qlog schema definitions SHOULD include a separate, length-indicating field for all fields of type hexstring they specify, see for example Section 10. This not only ensures the original length can always be retrieved, but also allows the omission of any raw value bytes of the field completely (e.g., out of privacy or security considerations).¶

To reduce overhead however and in the case the full raw value is logged, the extra length-indicating field can be left out. As such, tools MUST be able to deal with this situation and derive the length of the field from the raw value if no separate length-indicating field is present. The main possible permutations are shown by example in Figure 39.¶

// both the content's value and its length are present
// (length is redundant)
{
    "content_length": 5,
    "content": "051428abff"
}

// only the content value is present, indicating it
// represents the content's full value. The byte
// length is obtained by calculating content.length / 2
{
    "content": "051428abff"
}

// only the length is present, meaning the value
// was omitted
{
    "content_length": 5,
}

// both value and length are present, but the lengths
// do not match: the value was truncated to
// the first three bytes.
{
    "content_length": 5,
    "content": "051428"
}

11.5. Optimization of serialized data

Both the JSON and JSON-SEQ formatting options described above are serviceable in general small to medium scale (debugging) setups. However, these approaches tend to be relatively verbose, leading to larger file sizes. Additionally, generalized JSON(-SEQ) (de)serialization performance is typically (slightly) lower than that of more optimized and predictable formats. Both aspects present challenges to large scale setups, though they may still be practical to deploy; see [ANRW-2020]. JSON and JSON-SEQ compress very well using commonly-available algorithms such as GZIP or Brotli.¶

During the development of qlog, a multitude of alternative formatting and optimization options were assessed and the results are summarized on the qlog github repository.¶

Formal definition of additional qlog formats or encodings that use the optimization techniques described here, or any other optimization technique is left to future activity that can apply the following guidelines.¶

In order to help tools correctly parse and process serialized qlog, it is RECOMMENDED that new formats also define suitable file extensions and media types. This provides a clear signal and avoids the need to provide out-of-band information or to rely on heuristic fallbacks; see Section 13.¶

12.1. Set file output destination via an environment variable

To provide users control over where and how qlog files are created, two environment variables are defined. The first, QLOGFILE, indicates a full path to where an individual qlog file should be stored. This path MUST include the full file extension. The second, QLOGDIR, sets a general directory path in which qlog files should be placed. This path MUST include the directory separator character at the end.¶

In general, QLOGDIR should be preferred over QLOGFILE if an endpoint is prone to generate multiple qlog files. This can for example be the case for a QUIC server implementation that logs each QUIC connection in a separate qlog file. An alternative that uses QLOGFILE would be a QUIC server that logs all connections in a single file and uses the "group_id" field (Section 7.5) to allow post-hoc separation of events.¶

Implementations SHOULD provide support for QLOGDIR and MAY provide support for QLOGFILE.¶

When using QLOGDIR, it is up to the implementation to choose an appropriate naming scheme for the qlog files themselves. The chosen scheme will typically depend on the context or protocols used. For example, for QUIC, it is recommended to use the Original Destination Connection ID (ODCID), followed by the vantage point type of the logging endpoint. Examples of all options for QUIC are shown in Figure 40.¶

Command: QLOGFILE=/srv/qlogs/client.qlog quicclientbinary

Should result in the the quicclientbinary executable logging a
single qlog file named client.qlog in the /srv/qlogs directory.
This is for example useful in tests when the client sets up
just a single connection and then exits.

Command: QLOGDIR=/srv/qlogs/ quicserverbinary

Should result in the quicserverbinary executable generating
several logs files, one for each QUIC connection.
Given two QUIC connections, with ODCID values "abcde" and
"12345" respectively, this would result in two files:
/srv/qlogs/abcde_server.qlog
/srv/qlogs/12345_server.qlog

Command: QLOGFILE=/srv/qlogs/server.qlog quicserverbinary

Should result in the the quicserverbinary executable logging
a single qlog file named server.qlog in the /srv/qlogs directory.
Given that the server handled two QUIC connections before it was
shut down, with ODCID values "abcde" and "12345" respectively,
this would result in event instances in the qlog file being
tagged with the "group_id" field with values "abcde" and "12345".

14.1. Data at risk

qlog operators and implementers need to consider security and privacy risks when handling qlog data, including logging, storage, usage, and more. The considerations presented in this section may pose varying risks depending on the the data itself or its handling.¶

The following is a non-exhaustive list of example data types that could contain sensitive information that might allow identification or correlation of individual connections, endpoints, users or sessions across qlog or other data sources (e.g., captures of encrypted packets):¶

• IP addresses and transport protocol port numbers.¶
• Session, Connection, or User identifiers e.g., QUIC Connection IDs Section 9.5 of [RFC9000]).¶
• System-level information e.g., CPU, process, or thread identifiers.¶
• Stored State e.g., QUIC address validation and retry tokens, TLS session tickets, and HTTP cookies.¶
• TLS decryption keys, passwords, and HTTP-level API access or authorization tokens.¶
• High-resolution event timestamps or inter-event timings, event counts, packet sizes, and frame sizes.¶
• Full or partial raw packet and frame payloads that are encrypted.¶
• Full or partial raw packet and frame payloads that are plaintext e.g., HTTP Field values, HTTP response data, or TLS SNI field values.¶

14.2. Operational implications and recommendations

Operational considerations should focus on authorizing capture and access to logs. Logging of Internet protocols using qlog can be equivalent to the ability to store or read plaintext communications. Without a more detailed analysis, all of the security considerations of plaintext access apply.¶

It is recommended that qlog capture is subject to access control and auditing. These controls should support granular levels of information capture based on role and permissions (e.g., capture of more-sensitive data requires higher privileges).¶

It is recommended that access to stored qlogs is subject to access control and auditing.¶

End users might not understand the implications of qlog to security or privacy, and their environments might limit access control techniques. Implementations should make enabling qlog conspicuous (e.g., requiring clear and explicit actions to start a capture) and resistant to social engineering, automation, or drive-by attacks; for example, isolation or sandboxing of capture from other activities in the same process or component.¶

It is recommended that data retention policies are defined for the storage of qlog files.¶

It is recommended that qlog files are encrypted in transit and at rest.¶

14.3. Data minimization or anonymization

Applying data minimization or anonymization techniques to qlog might help address some security and privacy risks. However, removing or anonymizing data without sufficient care might not enhance privacy or security and could diminish the utility of qlog data.¶

Operators and implementers should balance the value of logged data with the potential risks of voluntary or involuntary disclosure to trusted or untrusted entities. Importantly, both the breadth and depth of the data needed to make it useful, as well as the definition of entities depend greatly on the intended use cases. For example, a research project might be tightly scoped, time bound, and require participants to explicitly opt in to having their data collected with the intention for this to be shared in a publication. Conversely, a server administrator might desire to collect telemetry, from users whom they have no relationship with, for continuing operational needs.¶

The most extreme form of minimization or anonymization is deleting a field, equivalent to not logging it. qlog implementations should offer fine-grained control for this on a per-use-case or per-connection basis.¶

Data can undergo anonymization, pseudonymization, permutation, truncation, re-encryption, or aggregation; see Appendix B of [DNS-PRIVACY] for techniques, especially regarding IP addresses. However, operators should be cautious because many anonymization methods have been shown to be insufficient to safeguard user privacy or identity, particularly with large or easily correlated data sets.¶

Operators should consider end user rights and preferences. Active user participation (as indicated by [RFC6973]) on a per-qlog basis is challenging but aligning qlog capture, storage, and removal with existing user preference and privacy controls is crucial. Operators should consider agressive approaches to deletion or aggregation.¶

The most sensitive data in qlog is typically contained in RawInfo type fields (see Section 10). Therefore, qlog users should exercise caution and limit the inclusion of such fields for all but the most stringent use cases.¶

Since draft-ietf-quic-qlog-main-schema-07:

• Added path and PathID (#336)¶
• Removed custom definition of uint64 type (#360, #388)¶
• ProtocolEventBody is now called ProtocolEventData (#352)¶
• Editorial changes (#364, #289, #353, #361, #362)¶

Since draft-ietf-quic-qlog-main-schema-06:

• Editorial reworking of the document (#331, #332)¶
• Updated IANA considerations section (#333)¶

Since draft-ietf-quic-qlog-main-schema-05:

• Updated qlog_version to 0.4 (due to breaking changes) (#314)¶
• Renamed 'transport' category to 'quic' (#302)¶
• Added 'system_info' field (#305)¶
• Removed 'summary' and 'configuration' fields (#308)¶
• Editorial and formatting changes (#298, #303, #304, #316, #320, #321, #322, #326, #328)¶

Since draft-ietf-quic-qlog-main-schema-04:

• Updated RawInfo definition and guidance (#243)¶

Since draft-ietf-quic-qlog-main-schema-03:

• Added security and privacy considerations discussion (#252)¶

Since draft-ietf-quic-qlog-main-schema-02:

• No changes - new draft to prevent expiration¶

Since draft-ietf-quic-qlog-main-schema-01:

• Change the data definition language from TypeScript to CDDL (#143)¶

Since draft-ietf-quic-qlog-main-schema-00:

• Changed the streaming serialization format from NDJSON to JSON Text Sequences (#172)¶
• Added Media Type definitions for various qlog formats (#158)¶
• Changed to semantic versioning¶

Since draft-marx-qlog-main-schema-draft-02:

• These changes were done in preparation of the adoption of the drafts by the QUIC working group (#137)¶
• Moved RawInfo, Importance, Generic events and Simulation events to this document.¶
• Added basic event definition guidelines¶
• Made protocol_type an array instead of a string (#146)¶

Since draft-marx-qlog-main-schema-01:

• Decoupled qlog from the JSON format and described a mapping instead (#89)¶

  • Data types are now specified in this document and proper definitions for fields were added in this format¶
  • 64-bit numbers can now be either strings or numbers, with a preference for numbers (#10)¶
  • binary blobs are now logged as lowercase hex strings (#39, #36)¶
  • added guidance to add length-specifiers for binary blobs (#102)¶
• Removed "time_units" from Configuration. All times are now in ms instead (#95)¶
• Removed the "event_fields" setup for a more straightforward JSON format (#101,#89)¶
• Added a streaming option using the NDJSON format (#109,#2,#106)¶
• Described optional optimization options for implementers (#30)¶
• Added QLOGDIR and QLOGFILE environment variables, clarified the .well-known URL usage (#26,#33,#51)¶
• Overall tightened up the text and added more examples¶

Since draft-marx-qlog-main-schema-00:

• All field names are now lowercase (e.g., category instead of CATEGORY)¶
• Triggers are now properties on the "data" field value, instead of separate field types (#23)¶
• group_ids in common_fields is now just also group_id¶