Real-time push and events
Zulip’s “events system” is the server-to-client push system that powers our real-time sync. This document explains how it works; to read an example of how a complete feature using this system works, check out the new application feature tutorial.
Any single-page web application like Zulip needs a story for how changes made by one client are synced to other clients, though having a good architecture for this is particularly important for a chat tool like Zulip, since the state is constantly changing. When we talk about clients, think a browser tab, mobile app, or API bot that needs to receive updates to the Zulip data. The simplest example is a new message being sent by one client; other clients must be notified in order to display the message. But a complete application like Zulip has dozens of different types of data that need to be synced to other clients, whether it be new streams, changes in a user’s name or avatar, settings changes, etc. In Zulip, we call these updates that need to be sent to other clients events.
An important thing to understand when designing such a system is that events need to be synced to every client that has a copy of the old data if one wants to avoid clients displaying inaccurate data to users. So if a user has two browser windows open and sends a message, every client controlled by that user as well as any recipients of the message, including both of those two browser windows, will receive that event. (Technically, we don’t need to send events to the client that triggered the change, but this approach saves a bunch of unnecessary duplicate UI update code, since the client making the change can just use the same code as every other client, maybe plus a little notification that the operation succeeded).
Architecturally, there are a few things needed to make a successful real-time sync system work:
Generation. Generating events when changes happen to data, and determining which users should receive each event.
Delivery. Efficiently delivering those events to interested clients, ideally in an exactly-once fashion.
UI updates. Updating the UI in the client once it has received events from the server.
This document discusses how Zulip solves the generation and delivery problems in a scalable, correct, and predictable way.
Zulip’s generation system is built around a Python function,
send_event(realm, event, users). It accepts the realm (used for
sharding), the event data structure (just a Python dictionary with
some keys and value;
type is always one of the keys but the rest
depends on the specific event) and a list of user IDs for the users
whose clients should receive the event. In special cases such as
message delivery, the list of users will instead be a list of dicts
mapping user IDs to user-specific data like whether that user was
mentioned in that message. The data passed to
send_event are simply
marshalled as JSON and placed in the
notify_tornado RabbitMQ queue
to be consumed by the delivery system.
Usually, this list of users is one of 3 things:
A single user (e.g. for user-level settings changes).
Everyone in the realm (e.g. for organization-level settings changes, like new realm emoji).
Everyone who would receive a given message (for messages, emoji reactions, message editing, etc.); i.e. the subscribers to a stream or the people on a private message thread.
It is the responsibility of the caller of
send_event to choose the
list of user IDs correctly. There can be security problems if e.g. an
event containing private message content is sent to the entire
organization. However, if an event isn’t sent to enough clients,
there will likely be user-visible real-time sync bugs.
Most of the hard work in event generation is about defining consistent event dictionaries that are clear, readable, will be useful to the wide range of possible clients, and make it easy for developers.
Zulip’s event delivery (real-time push) system is based on Tornado,
which is ideal for handling a large number of open requests. Details
on Tornado are available in the
architecture overview, but in short it
is good at holding open a large number of connections for a long time.
The complete system is about 2000 lines of code in
Zulip’s event delivery system is based on “long-polling”; basically
GET /json/events calls to the server, and the server
doesn’t respond to the request until it has an event to deliver to the
client. This approach is reasonably efficient and works everywhere
(unlike websockets, which have a decreasing but nonzero level of
client compatibility problems).
For each connected client, the event queue server maintains an
event queue, which contains any events that are to be delivered to
that client which have not yet been acknowledged by that client.
Ignoring the subtle details around error handling, the protocol is
pretty simple; when a client does a
GET /json/events call, the
server checks if there are any events in the queue. If there are, it
returns the events immediately. If there aren’t, it records that
queue as having a waiting client (often called a
handler in the
When it pulls an event off the
notify_tornado RabbitMQ queue, it
simply delivers the event to each queue associated with one of the
target users. If the queue has a waiting client, it breaks the
long-poll connection by returning an HTTP response to the waiting
client request. If there is no waiting client, it simply pushes the
event onto the queue.
When starting up, each client makes a
POST /json/register to the
server, which creates a new event queue for that client and returns the
queue_id as well as an initial
last_event_id to the client (it can
also, optionally, fetch the initial data to save an RTT and avoid
races; see the below section on initial data fetches for details on
why this is useful). Once the event queue is registered, the client
can just do an infinite loop calling
GET /json/events with those
last_event_id each time to acknowledge any
events it has received (see
call_on_each_event in the
Zulip Python API bindings for a complete example
implementation). When handling each
GET /json/events request, the
queue server can safely delete any events that have an event ID less
than or equal to the client’s
last_event_id (event IDs are just a
counter for the events a given queue has received.)
If network failures were impossible, the
last_event_id parameter in
the protocol would not be required, but it is important for enabling
exactly-once delivery in the presence of potential failures. (Without
it, the queue server would have to delete events from the queue as
soon as it attempted to send them to the client; if that specific HTTP
response didn’t reach the client due to a network TCP failure, then
those events could be lost).
The queue servers are a very high-traffic system, processing at a minimum one request for every message delivered to every Zulip client. Additionally, as a workaround for low-quality NAT servers that kill HTTP connections that are open without activity for more than 60s, the queue servers also send a heartbeat event to each queue at least once every 45s or so (if no other events have arrived in the meantime).
To avoid a large memory and other resource leak, the queues are garbage collected after (by default) 10 minutes of inactivity from a client, under the theory that the client has likely gone off the Internet (or no longer exists) access; this happens constantly. If the client returns, it will receive a “queue not found” error when requesting events; its handler for this case should just restart the client / reload the browser so that it refetches initial data the same way it would on startup. Since clients have to implement their startup process anyway, this approach adds minimal technical complexity to clients. A nice side effect is that if the event queue server (which stores queues in memory) were to crash and lose its data, clients would recover, just as if they had lost Internet access briefly (there is some DoS risk to manage, though).
Note that the garbage-collection system has hooks that are important for the implementation of notifications.
(The event queue server is designed to save any event queues to disk and reload them when the server is restarted, and catches exceptions carefully, so such incidents are very rare, but it’s nice to have a design that handles them without leaving broken out-of-date clients anyway).
The initial data fetch
When a client starts up, it usually wants to get 2 things from the server:
The “current state” of various pieces of data, e.g. the current settings, set of users in the organization (for typeahead), stream, messages, etc. (aka the “initial state”).
A subscription to receive updates to those data when they are changed by a client (aka an event queue).
Ideally, one would get those two things atomically, i.e. if some other user changes their name, either the name change happens after the fetch (and thus the old name is in the initial state and there will be an event in the queue for the name change) or before (the new name is in the initial state, and there is no event for that name change in the queue).
Achieving this atomicity goals means we save a huge amount of work that the N clients for Zulip don’t need to worry about a wide range of potential rare and hard to reproduce race conditions; we just have to implement things correctly once in the Zulip server.
This is quite challenging to do technically, because fetching the
initial state for a complex web application like Zulip might involve
dozens of queries to the database, caches, etc. over the course of
100ms or more, and it is thus nearly impossible to do all of those
things together atomically. So instead, we use a more complicated
algorithm that can produce the atomic result from non-atomic
subroutines. Here’s how it works when you make a
request; the logic is in
zerver/lib/events.py. The request is directly handled by Django:
Django makes an HTTP request to Tornado, requesting that a new event queue be created, and records its queue ID.
Django does all the various database/cache/etc. queries to fetch the data, non-atomically, from the various data sources (see the
Django makes a second HTTP request to Tornado, requesting any events that had been added to the Tornado event queue since it was created.
Finally, Django “applies” the events (see the
apply_eventsfunction) to the initial state that it fetched. E.g. for a name change event, it finds the user data in the
realm_userdata structure, and updates it to have the new name.
The design above achieves everything we desire, at the cost that we need to
write a correct
apply_events function. This is a difficult function to
implement correctly, because the situations that it handles almost never
happen (being race conditions) during manual testing. Fortunately, we have
a protocol for testing
apply_events in our automated backend tests.
Once you are completely confident that an “action function” works correctly
in terms of “normal” operation (which typically involves writing a
full-stack test for the corresponding POST/GET operation), then you will be
ready to write a test in
The actual code for a
test_events test can be quite concise:
def test_default_streams_events(self) -> None: stream = get_stream("Scotland", self.user_profile.realm) events = self.verify_action(lambda: do_add_default_stream(stream)) check_default_streams("events", events) # (some details omitted)
The real trick is debugging these tests.
The test example above has three things going on:
Set up some data (
verify_actionwith an action function (
Use a schema checker to validate data (
All the heavy lifting that pertains to
apply_events happens within the
verify_action, which is a test helper in the
verify_action function simulates the possible race condition in
order to verify that the
apply_events logic works correctly in the
context of some action function. To use our concrete example above,
we are seeing that applying the events from the
do_remove_default_stream action inside of
apply_events to a stale
copy of your state results in the same state dictionary as doing the
action and then fetching a fresh copy of the state.
verify_action does the following:
fetch_initial_state_datato get the current state.
Call the action function (e.g.
Capture the events generated by the action function.
Check the events generated are documented in the OpenAPI schema defined in
apply_events(state, events), to get the resulting “hybrid state”.
fetch_initial_state_dataagain to get the “normal state”.
Compare the two results.
In the event that you wrote the
apply_events logic correctly the
first time, then the two states will be identical, and the
verify_action call will succeed and return the events that came from
Often you will get the
apply_events logic wrong at first, which will
verify_action to fail. To help you debug, it will print a diff
between the “hybrid state” and the “normal state” obtained from
fetch_initial_state_data after the changes. If you encounter a
diff like this, you may be in for a challenging debugging exercise. It
will be helpful to re-read this documentation to understand the rationale
apply_events function. It may also be helpful to read the code
verify_action itself. Finally, you may want to ask for help on chat.
Before we move on to the next step, it’s worth noting that
only has one required parameter, which is the action function. We
typically express the action function as a lambda, so that we
can pass in arguments:
events = self.verify_action(lambda: do_add_default_stream(stream))
There are some notable optional parameters for
state_change_expectedmust be set to
Falseif your action doesn’t actually require state changes for some reason; otherwise,
verify_actionwill complain that your test doesn’t really exercise any
apply_eventslogic. Typing notifications (which are ephemeral) are a common place where we use this.
verify_actionhow many events the
hamletuser will receive after the action (the default is 1).
parameters such as
slim_presenceget passed along to
fetch_initial_state_data(and it’s important to test both boolean values of these parameters for relevant actions).
For advanced use cases of
verify_action, we highly recommend reading
the code itself in
test_events.py system has two forms of schema checking. The
first is verifying that you’ve updated the GET /events API
documentation to document your new
event’s format for benefit of the developers of Zulip’s mobile app,
terminal app, and other API clients. See the API documentation
docs for details on the OpenAPI
The second is higher-detail check inside
test_events that this
specific test generated the expected series of events. Let’s look at
the last line of our example test snippet:
# ... events = self.verify_action(lambda: do_add_default_stream(stream)) check_default_streams("events", events)
We have discussed
verify_action in some detail, and you will
note that it returns the actual events generated by the action
function. It is part of our test discipline in
verify that the events are formatted in a predictable way.
Ideally, we would test that events match the exact data that we
expect, but it can be difficult to do this due to unpredictable
things like database ids. So instead, we just verify the “schema”
of the event(s). We use a schema checker like
to validate the types of the data.
If you are creating a new event format, then you will have to
write your own schema checker in
event_schema.py. Here is
the example relevant to our example:
default_streams_event = event_dict_type( required_keys=[ ("type", Equals("default_streams")), ("default_streams", ListType(DictType(basic_stream_fields))), ] ) check_default_streams = make_checker(default_streams_event)
basic_stream_fields is not shown in these docs. The
best way to understand how to write schema checkers is to read
event_schema.py. There is a large block comment at the top of
the file, and then you can skim the rest of the file to see the
When you create a new schema checker for a new event, you not only
test_events test more rigorous, you also allow our other
tools to use the same schema checker to validate event formats in our
node test fixtures and our OpenAPI documentation.
Once you’ve completed backend testing, be sure to add an example event
frontend_tests/node_tests/lib/events.js, a test of the
server_events_dispatch.js code for that event in
frontend_tests/node_tests/dispatch.js, and verify your example
against the two versions of the schema that you declared above using
The final detail we need to ensure that
apply_events always works
correctly is to make sure that we have relevant tests for
every event type that can be generated by Zulip. This can be tested
test-backend --coverage BaseAction and then
checking that all the calls to
send_event are covered. Someday
we’ll add automation that verifies this directly by inspecting the
In the Zulip web app, the data returned by the
register API is
available via the
One exception to the protocol described in the last section is the actual messages. Because Zulip clients usually fetch them in a separate AJAX call after the rest of the site is loaded, we don’t need them to be included in the initial state data. To handle those correctly, clients are responsible for discarding events related to messages that the client has not yet fetched.
Additionally, see the main documentation on sending messages.
When changing the format of events sent into Tornado, it’s important to make sure we handle backwards-compatibility properly.
If we’re adding a new event type or new fields to an existing event type, we just need to carefully document the changes in the API documentation, being careful to bump
API_FEATURE_LEVELand include a
**Changes**entry in the updated
GET /eventsAPI documentation. It’s also a good idea to and open issues with the mobile and terminal projects to notify them.
If we’re making changes that could confuse existing client app logic that parses events (E.g. changing the type/meaning of an existing field, or removing a field), we need to be very careful, since Zulip supports old clients connecting to a modern server. See our release lifecycle documentation for more details on the policy. Our technical solution is to add a
client_capabilitiesflag for the new format and have the code continue sending data in the old format for clients that don’t declare support for the new capability.
bulk_message_deletionis a good example to crib from. (A few years later, we’ll make the client capability required and remove support for not having it).
For most event types, Tornado just passes the event through transparently, and
event_queue.pyrequires no changes.
However, when changing the format of data used by Tornado code, like renaming the
messageevents, we need to be careful, because pre-upgrade events may be present in Tornado’s queue when we upgrade Tornado. So it’s essential to write logic in
event_queue.pyto translate the old format into the new format, or Tornado may crash when upgrading past the relevant commit. We attempt to contain that sort of logic in the
from_dictfunction (which is used for changing event queue formats) and
client_capabilitiesconditionals (E.g. in
process_deletion_event). Compatibility code not related to a
client_capabilitiesentry should be marked with a
# TODO/compatibility: ...comment noting when it can be safely deleted; we grep for these comments entries during major releases.
Schema changes are a sensitive operation, and like with database schema changes, it’s critical to do thoughtful manual testing. E.g. run the mobile app against your test server and verify it handles the new event properly, or arrange for your new Tornado code to actually process a pre-upgrade event and verify via the browser console what came out.