Summary:
Previously we track them since we thought we can get some additional
RTT samples. But these are bad RTT samples since peer can delays the acking of
pure acks. Now we no longer trust such RTT samples, there is no reason to keep
tracking pure ack packets.
Reviewed By: mjoras
Differential Revision: D18946081
fbshipit-source-id: 0a92d88e709edf8475d67791ba064c3e8b7f627a
Summary:
In the current client code we read one packet, go back to epoll, and then read
another packet. This is not very efficient.
This changes it so that we can read multiple packets in one go from an epoll
callback.
This only performs changes on the client
Reviewed By: mjoras
Differential Revision: D18797962
fbshipit-source-id: 81be82111064ade4fe3a07b1d9d3d01e180f29f5
Summary: The state machine logic is quite abstruse, this modifies it to make it more readable.
Reviewed By: siyengar
Differential Revision: D18488301
fbshipit-source-id: c6fd52973880931e34904713e8b147f56d0c4629
Summary:
F14 should be faster and have lower memory urilization for near-empty sets and maps. For most H3 connections these are mosotly going to be near-empty, so CPU wins will likely be minimal.
For usecases that have extremely high numbers of streams, there are likely going to be CPU wins.
Reviewed By: yangchi
Differential Revision: D18484047
fbshipit-source-id: 7f5616d6d6c8651ca5b03468d7d8895d1f51cb53
Summary:
QLogConnectionMigrationEvent:
Allow observing client-side ConnectionMigration attempts (replacing the
socket), and observing the server-side changing the peer address it
is writing to.
QLogPathValidationEvent:
Allow observing successful/failed path validation attempts.
Success is considered as a correct PathResponse being returned.
A Failure is only published on the timeout expiring, not an invalid
PathChallenge frame being returned (we do not cancel this).
There are already QlogEvents for PathChallenge/PathResponse that
can be observed.
Reviewed By: JunqiWang
Differential Revision: D18340999
fbshipit-source-id: 512108f82a6e082021c0bd3254f108c128b17ba3
Summary: inspect PN, and when it reaches 2^62 -2 trigger a transport close through a pending event.
Reviewed By: yangchi
Differential Revision: D18239661
fbshipit-source-id: 1a218678099016693149e12ff121e2a39b95aecc
Summary: Enable pacing when CC is changed to BBR
Reviewed By: yangchi
Differential Revision: D18231888
fbshipit-source-id: a54b6313d089c2ae24ce6bf6ee56c4fe0d6b4722
Summary:
We currently pause all producers if the sum of the egress buffers of all transactions exceeds the write buffer limit. This turns out to be deterimental to prioritization.
Now, we pass the underlying transport pause state or connection flow control state back to the handlers. The previous diff in this stack introduces a per-stream buffer limit (64kb default). To limit total session buffer size, limit the number of concurrent streams or lower the per-stream limit.
Reviewed By: lnicco
Differential Revision: D17097138
fbshipit-source-id: 9025c5be8b318963311c3aaad9ee9a03c0e2265e
Summary: Use the custom variant type for errors.
Reviewed By: yangchi
Differential Revision: D17826935
fbshipit-source-id: 2bf0c3e1cc8ca84b504d201fd6c2a4266878b715
Summary:
CC Type is not exposed anywhere else. This could be the source of truth and will be useful for logging.
Also added a helper to convert the enum to string.
Reviewed By: yangchi
Differential Revision: D17566664
fbshipit-source-id: 1e0e887a7c23617b174b240f5c636f6dcdfd42c4
Summary:
Currently the lower bound of ack timeout is kMinAckTimeout which is
10ms. This diff changes to use timer's tick interval as lower bound. For the
timer we use today (the default HHWheelTimer), this is a no-op change since the
tick interval is also 10ms.
Reviewed By: mjoras
Differential Revision: D17421600
fbshipit-source-id: 073ac6a8e5d84dbdfc00e8e95ff13be26adb1684
Summary:
Implement sending stream limit updates in a windowed fashion, so that as a peer exhausts its streams we will grant it additional credit. This is implemented by having the stream manager check if an update is needed on removing streams, and the api layer potentially sending an update after it initiates the check for closed streams.
This also makes some driveby changes to use `std::lower_bound` instead of `std::find` for the sorted collections in the stream manager.
Reviewed By: yangchi
Differential Revision: D16808229
fbshipit-source-id: f6e3460d43e4d165e362164be00c0cec27cf1e79
Summary:
we were not setting the MSS.
populate it from the transport, compute the cwnd appropriately and check the netquality header
Reviewed By: yangchi
Differential Revision: D17385975
fbshipit-source-id: 3666728c48d3fc7a65827d526fbdd546449b2e1f
Summary:
This is important in production, but not necessarily for standalone clients.
The transport drain feature is functionally similar to tcp TIME_WAIT, with the
difference that with a userspace transport implementation that forces a client
application to wait for that timeout before it can shutdown
Reviewed By: mjoras
Differential Revision: D17271443
fbshipit-source-id: db5d9e79f175aa86cd9ca8a8a2181b5e80c1d0e8
Summary:
When apps interested in such info retrieve it, it would be bad if we
update the value and mess up the timer drift. So this diff adds a pure getter
function.
Reviewed By: mjoras
Differential Revision: D16918673
fbshipit-source-id: c4954b2f7e3a1d53ec0e6491ef1a25e8f53f1744
Summary:
Use the new Pacer interface in the transport where we currently
directly use CongestinoController interrace for paciner related APIs.
Reviewed By: mjoras
Differential Revision: D16918672
fbshipit-source-id: 410f72d567e71bdd3279e344f6e9abf5e132518e
Summary:
Second try of this diff. This time after the fix of uninitizlied
default constructed std::chrono::duration values in BBR and Copa.
Currently each conegstion controller maintains this value while
TransportSettings also has this value. But currently there is no way they are
in sync. The timer uses the value in TransportSettings. So the value maintained
by each congestion controller isn't used except for tracing, but that also make
tracing wrong.
Reviewed By: oesh
Differential Revision: D16985568
fbshipit-source-id: 5fbd9fee4deec5177b92c733a159e0ba9bfd2289
Summary:
Mostly need a bunch of folly::assume_unreachable() (sometimes
replacing __builtin_unreachable()). One place using designated initializers
that had to get commented out. Some headers replaced with folly portability ones.
Reviewed By: mzlee, phoad
Differential Revision: D16970520
fbshipit-source-id: 1b8a36ecb1975e2dc0869b66c4ea5439baf7575d
Summary:
Currently each conegstion controller maintains this value while
TransportSettings also has this value. But currently there is no way they are
in sync. The timer uses the value in TransportSettings. So the value maintained
by each congestion controller isn't used except for tracing, but that also make
tracing wrong.
Reviewed By: mjoras
Differential Revision: D16772744
fbshipit-source-id: 71800e5dc1f9fce8df4b1f72ce5aa5cfd15fbc89
Summary: Update qlog format to be more complete. Adding the summary section (with extra fields like title, description, etc). This diff is just to make the format more on par with Robin's schema.
Reviewed By: mjoras
Differential Revision: D16499808
fbshipit-source-id: 56cfbb95404f7e3c6638bffda18b53f7d83048a1
Summary: Add transportStateUpdate event so it can be part of qlog.
Reviewed By: mjoras
Differential Revision: D16342467
fbshipit-source-id: 109189275d44996850b82646bab4a733a3a4c7a1
Summary:
Previously we tried to put this API in connectionCallback, but that has
some lifecycle problem. So we decided to make this API independent of mvfst's
state. Application (e.g. HTTP/3, QPACK) will have to either make the 2 functions
stateless or maintain the lifecycle of any captured objects.
Reviewed By: afrind, lnicco
Differential Revision: D16074471
fbshipit-source-id: b0a3baa6870bee56372797754972ab5c2c1f2232
Summary:
This diff adds a DebugState in the QuicConnectionState to track the
reason we schedule transport WriteLooper to run, the reason we end up not
writing and the number of times such write happens consecutively. And when it
reaches a predefined limit, we trigger a callback on a loop detector interface.
Reviewed By: kvtsoy
Differential Revision: D15433881
fbshipit-source-id: d903342c00bc4dccf8d7320726368d23295bec66
Summary: Add TransportSummary event to Quic, so it can be logged as part of qlog.
Reviewed By: mjoras
Differential Revision: D16116127
fbshipit-source-id: ddfecac5b5452fe9b2e3df87b152561eab903188
Summary: This is useful information for the application to know about
Reviewed By: yangchi
Differential Revision: D15599964
fbshipit-source-id: da451f46238f68fe701d6f433701c569cfe2f47d
Summary:
When a write callback x is already defined for a stream
and notifyPendingWriteOnStream is called with the same callback x,
it is safe to not return an error.
Building on master is currently broken due to 2040a13e40 (diff-ddd35d436da215e41b89fd5afe4acebbR11), but this commit builds when rebased before that.
udippant
Pull Request resolved: https://github.com/facebookincubator/mvfst/pull/18
Reviewed By: yangchi
Differential Revision: D15562893
Pulled By: afrind
fbshipit-source-id: 44e62bb9bb0002508257acdc419f20bcb973ca64
Summary: bytesAcked is a very useful stat for apps to know. Bytes written only tells me that data was written to socket but it may not have been acked.
Differential Revision: D15507105
fbshipit-source-id: 656ee45418a78c7e05d8e7979afa2c9a41976e85
Summary:
Current app-limited is defined as the connection doesn't have a no
non-control stream. I'm changing this to be app-idle. And app-limited will be
changed to a state in connection where app isn't sending bytes fast enough to
keep congestion controller probing higher bandwidth.
Reviewed By: mjoras
Differential Revision: D15456199
fbshipit-source-id: e084bc133284ae37ee6da8ebb3c12f505011521e
Summary: Just to make it consistent with other var names in quic
Reviewed By: siyengar
Differential Revision: D15415528
fbshipit-source-id: 917159d6ccf86cf34612386d128d6b6d5bf62de6
Summary:
Peek callback wasn't raised on data arrival. The fix is to:
* always update peek looper before read looper
* always update `peekableList` together with `readableList`
Differential Revision: D14757064
fbshipit-source-id: bb1e43762f54d024050eced425351f03a0cffd9f
Summary:
Using FOLLY_MOBILE macro to be able to exclude
quic trace generation on mobile
Reviewed By: mjoras
Differential Revision: D14609755
fbshipit-source-id: 2d5d624703a5b58c28cb533e3c2cc20d3a5c3d45
Summary:
Draft-19 onwards effectively punted version negotiation to QUICv2. Now receiving version negotiation on clients is treated as an immediate termination of the connection.
The transport parameter format has also changed to no longer include any reference to the QUIC version. To avoid us (Facebook) having to turn off QUIC traffic in production, our server needs to be able to parse these transport parameters from our older clients. To achieve this when parsing the transport parameters we will, as a temporary measure, check for the Facebook QUIC version to determine which transport parameter format we are parsing. Luckily for us the version we chose maps nicely to an implausible length for the transport parameters (0xface).
Note that this diff still has the client send the old transport parameter format, so that the rollout can be staged.
Reviewed By: yangchi
Differential Revision: D15203481
fbshipit-source-id: dfaaddc3acc76434461b04430b82a0902138c060
Summary:
Use find instead of count. We just need to know whether one exists, count will go through
all the elements instead which is a waste
Reviewed By: sharma95
Differential Revision: D15270589
fbshipit-source-id: f85780f524f6534b0b240162b60b7c7d72c0df85
Summary: To conform to the current specs.
Reviewed By: mjoras
Differential Revision: D15215018
fbshipit-source-id: 4dcc495aea1cd7cebf2bc84f7367cb2e4a55df19
Summary:
Previously updateAppLimited was not accounting for peer created streams.
This moves app limited accounting into the stream manager
This also makes a subtle change to the applimited callback. We assume that
we start off as not app limited now, so if we create a stream we will not call
the congestion controller app limited callback.
Reviewed By: mjoras
Differential Revision: D15166114
fbshipit-source-id: 1a8d573533861f53bb1bd9fdc605dfefe68902dc
Summary: The previous diff removed automatically sending a rst stream when receiving a reset stream. This adds a helper to do that from the application layer.
Reviewed By: mjoras, pkir
Differential Revision: D15208997
fbshipit-source-id: b8d3898e85537e97df61cc66854299698a48c825
Summary: This is step 1 for removing reset on reset, since the send side may need to transition to waiting for a reset ack while the read side is an any state.
Reviewed By: lnicco
Differential Revision: D15075849
fbshipit-source-id: 1e094942a8a1ca9a01d4161cd6309b4136a9cfbf
