sliding-sync/tests-e2e/num_live_test.go

200 lines
5.7 KiB
Go
Raw Normal View History

add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
package syncv3_test
import (
"fmt"
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
"testing"
"time"
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
2023-10-11 12:23:46 +01:00
"github.com/matrix-org/complement/b"
"github.com/matrix-org/sliding-sync/sync3"
"github.com/matrix-org/sliding-sync/testutils/m"
"github.com/tidwall/gjson"
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
)
func TestNumLive(t *testing.T) {
alice := registerNewUser(t)
bob := registerNewUser(t)
2023-10-11 12:23:46 +01:00
roomID := alice.MustCreateRoom(t, map[string]interface{}{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
"preset": "public_chat",
})
bob.JoinRoom(t, roomID, nil)
2023-10-11 12:23:46 +01:00
eventID := alice.SendEventSynced(t, roomID, b.Event{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
Type: "m.room.message",
Content: map[string]interface{}{
"msgtype": "m.text",
"body": "Hello World!!!!",
},
})
// initial syncs -> no live events
res := bob.SlidingSync(t, sync3.Request{
RoomSubscriptions: map[string]sync3.RoomSubscription{
roomID: {
TimelineLimit: 1,
},
},
})
m.MatchResponse(t, res, m.MatchRoomSubscriptionsStrict(
map[string][]m.RoomMatcher{
roomID: {
MatchRoomTimelineContains(Event{ID: eventID}),
m.MatchNumLive(0),
},
},
))
// live event -> 1 num live
2023-10-11 12:23:46 +01:00
eventID2 := alice.SendEventSynced(t, roomID, b.Event{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
Type: "m.room.message",
Content: map[string]interface{}{
"msgtype": "m.text",
"body": "Hello World!!!! 2",
},
})
res = bob.SlidingSyncUntilEventID(t, res.Pos, roomID, eventID2)
m.MatchResponse(t, res, m.MatchRoomSubscriptionsStrict(
map[string][]m.RoomMatcher{
roomID: {
MatchRoomTimelineContains(Event{ID: eventID2}),
m.MatchNumLive(1),
},
},
))
// from fresh => 0 num live
res = bob.SlidingSyncUntilEventID(t, "", roomID, eventID2)
m.MatchResponse(t, res, m.MatchRoomSubscriptionsStrict(
map[string][]m.RoomMatcher{
roomID: {
MatchRoomTimelineContains(Event{ID: eventID2}),
m.MatchNumLive(0),
},
},
))
// now the big one -> 3 rooms, ask for top 2, bump 3rd room to top twice -> num_live=2
2023-10-11 12:23:46 +01:00
roomID2 := alice.MustCreateRoom(t, map[string]interface{}{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
"preset": "public_chat",
})
bob.JoinRoom(t, roomID2, nil)
2023-10-11 12:23:46 +01:00
roomID3 := alice.MustCreateRoom(t, map[string]interface{}{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
"preset": "public_chat",
})
bob.JoinRoom(t, roomID3, nil)
// let the proxy become aware of these joins
alice.SlidingSyncUntilMembership(t, "", roomID3, bob, "join")
// at this point, roomID is at the bottom, check.
res = bob.SlidingSync(t, sync3.Request{
Lists: map[string]sync3.RequestList{
"a": {
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
Ranges: sync3.SliceRanges{{0, 1}}, // top 2 rooms
Sort: []string{sync3.SortByRecency},
RoomSubscription: sync3.RoomSubscription{
TimelineLimit: 2,
},
},
},
})
m.MatchResponse(t, res, m.MatchList("a", m.MatchV3Ops(
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
m.MatchV3SyncOp(0, 1, []string{roomID3, roomID2}),
)))
// now send 2 live events into roomID to bump it to the top
2023-10-11 12:23:46 +01:00
eventID3 := alice.SendEventSynced(t, roomID, b.Event{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
Type: "m.room.message",
Content: map[string]interface{}{
"msgtype": "m.text",
"body": "bump 1",
},
})
2023-10-11 12:23:46 +01:00
eventID4 := alice.SendEventSynced(t, roomID, b.Event{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
Type: "m.room.message",
Content: map[string]interface{}{
"msgtype": "m.text",
"body": "bump 2",
},
})
// wait for the proxy to get it
alice.SlidingSyncUntilEventID(t, "", roomID, eventID4)
// now syncing with bob should see 2 live events
res = bob.SlidingSync(t, sync3.Request{
Lists: map[string]sync3.RequestList{
"a": {
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
Ranges: sync3.SliceRanges{{0, 1}}, // top 2 rooms
},
},
}, WithPos(res.Pos))
m.MatchResponse(t, res, m.MatchRoomSubscriptionsStrict(map[string][]m.RoomMatcher{
roomID: {
m.MatchNumLive(2),
MatchRoomTimeline([]Event{{ID: eventID2}, {ID: eventID3}, {ID: eventID4}}),
},
}))
}
// Test that if you constantly change req params, we still see live traffic. It does this by:
// - Creating 11 rooms.
// - Hitting /sync with a range [0,1] then [0,2] then [0,3]. Each time this causes a new room to be returned.
// - Interleaving each /sync request with genuine events sent into a room.
// - ensuring we see the genuine events by the time we finish.
func TestReqParamStarvation(t *testing.T) {
alice := registerNewUser(t)
bob := registerNewUser(t)
2023-10-11 12:23:46 +01:00
roomID := alice.MustCreateRoom(t, map[string]interface{}{
"preset": "public_chat",
})
numOtherRooms := 10
for i := 0; i < numOtherRooms; i++ {
2023-10-11 12:23:46 +01:00
bob.MustCreateRoom(t, map[string]interface{}{
"preset": "public_chat",
})
}
bob.JoinRoom(t, roomID, nil)
res := bob.SlidingSyncUntilMembership(t, "", roomID, bob, "join")
wantEventIDs := make(map[string]bool)
for i := 0; i < numOtherRooms; i++ {
res = bob.SlidingSync(t, sync3.Request{
Lists: map[string]sync3.RequestList{
"a": {
Ranges: sync3.SliceRanges{{0, int64(i)}}, // [0,0], [0,1], ... [0,9]
},
},
}, WithPos(res.Pos))
// mark off any event we see in wantEventIDs
for _, r := range res.Rooms {
for _, ev := range r.Timeline {
gotEventID := gjson.GetBytes(ev, "event_id").Str
wantEventIDs[gotEventID] = false
}
}
// send an event in the first few syncs to add to wantEventIDs
// We do this for the first few /syncs and don't dictate which response they should arrive
// in, as we do not know and cannot force the proxy to deliver the event in a particular response.
if i < 3 {
2023-10-11 12:23:46 +01:00
eventID := alice.SendEventSynced(t, roomID, b.Event{
Type: "m.room.message",
Content: map[string]interface{}{
"msgtype": "m.text",
"body": fmt.Sprintf("msg %d", i),
},
})
wantEventIDs[eventID] = true
}
// it's possible the proxy won't see this event before the next /sync
// and that is the reason why we don't send it, as opposed to starvation.
// To try to counter this, sleep a bit. This is why we sleep on every cycle and
// why we send the events early on.
time.Sleep(50 * time.Millisecond)
}
// at this point wantEventIDs should all have false values if we got the events
for evID, unseen := range wantEventIDs {
if unseen {
t.Errorf("failed to see event %v", evID)
}
}
}