dendrite/roomserver/internal/api.go

227 lines
6.6 KiB
Go
Raw Normal View History

package internal
import (
"context"
"github.com/getsentry/sentry-go"
asAPI "github.com/matrix-org/dendrite/appservice/api"
fsAPI "github.com/matrix-org/dendrite/federationapi/api"
"github.com/matrix-org/dendrite/internal/caching"
"github.com/matrix-org/dendrite/roomserver/acls"
"github.com/matrix-org/dendrite/roomserver/api"
"github.com/matrix-org/dendrite/roomserver/internal/input"
"github.com/matrix-org/dendrite/roomserver/internal/perform"
"github.com/matrix-org/dendrite/roomserver/internal/query"
"github.com/matrix-org/dendrite/roomserver/storage"
"github.com/matrix-org/dendrite/setup/config"
"github.com/matrix-org/dendrite/setup/process"
userapi "github.com/matrix-org/dendrite/userapi/api"
"github.com/matrix-org/gomatrixserverlib"
Add NATS JetStream support (#1866) * Add NATS JetStream support Update shopify/sarama * Fix addresses * Don't change Addresses in Defaults * Update saramajetstream * Add missing error check Keep typing events for at least one minute * Use all configured NATS addresses * Update saramajetstream * Try setting up with NATS * Make sure NATS uses own persistent directory (TODO: make this configurable) * Update go.mod/go.sum * Jetstream package * Various other refactoring * Build fixes * Config tweaks, make random jetstream storage path for CI * Disable interest policies * Try to sane default on jetstream base path * Try to use in-memory for CI * Restore storage/retention * Update nats.go dependency * Adapt changes to config * Remove unneeded TopicFor * Dep update * Revert "Remove unneeded TopicFor" This reverts commit f5a4e4a339b6f94ec215778dca22204adaa893d1. * Revert changes made to streams * Fix build problems * Update nats-server * Update go.mod/go.sum * Roomserver input API queuing using NATS * Fix topic naming * Prometheus metrics * More refactoring to remove saramajetstream * Add missing topic * Don't try to populate map that doesn't exist * Roomserver output topic * Update go.mod/go.sum * Message acknowledgements * Ack tweaks * Try to resume transaction re-sends * Try to resume transaction re-sends * Update to matrix-org/gomatrixserverlib@91dadfb * Remove internal.PartitionStorer from components that don't consume keychanges * Try to reduce re-allocations a bit in resolveConflictsV2 * Tweak delivery options on RS input * Publish send-to-device messages into correct JetStream subject * Async and sync roomserver input * Update dendrite-config.yaml * Remove roomserver tests for now (they need rewriting) * Remove roomserver test again (was merged back in) * Update documentation * Docker updates * More Docker updates * Update Docker readme again * Fix lint issues * Send final event in `processEvent` synchronously (since this might stop Sytest from being so upset) * Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that * Go 1.16 instead of Go 1.13 for upgrade tests and Complement * Revert "Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that" This reverts commit 368675283fc44501f227639811bdb16dd5deef8c. * Don't report any errors on `/send` to see what fun that creates * Fix panics on closed channel sends * Enforce state key matches sender * Do the same for leave * Various tweaks to make tests happier Squashed commit of the following: commit 13f9028e7a63662759ce7c55504a9d2423058668 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 15:47:14 2022 +0000 Do the same for leave commit e6be7f05c349fafbdddfe818337a17a60c867be1 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 15:33:42 2022 +0000 Enforce state key matches sender commit 85ede6d64bf10ce9b91cdd6d80f87350ee55242f Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 14:07:04 2022 +0000 Fix panics on closed channel sends commit 9755494a98bed62450f8001d8128e40481d27e15 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 13:38:22 2022 +0000 Don't report any errors on `/send` to see what fun that creates commit 3bb4f87b5dd56882febb4db5621db484c8789b7c Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 13:00:26 2022 +0000 Revert "Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that" This reverts commit 368675283fc44501f227639811bdb16dd5deef8c. commit fe2673ed7be9559eaca134424e403a4faca100b0 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 12:09:34 2022 +0000 Go 1.16 instead of Go 1.13 for upgrade tests and Complement commit 368675283fc44501f227639811bdb16dd5deef8c Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 11:51:45 2022 +0000 Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that commit b028dfc08577bcf52e6cb498026e15fa5d46d07c Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 10:29:08 2022 +0000 Send final event in `processEvent` synchronously (since this might stop Sytest from being so upset) * Merge in NATS Server v2.6.6 and nats.go v1.13 into the in-process connection fork * Add `jetstream.WithJetStreamMessage` to make ack/nak-ing less messy, use process context in consumers * Fix consumer component name in federation API * Add comment explaining where streams are defined * Tweaks to roomserver input with comments * Finish that sentence that I apparently forgot to finish in INSTALL.md * Bump version number of config to 2 * Add comments around asynchronous sends to roomserver in processEventWithMissingState * More useful error message when the config version does not match * Set version in generate-config * Fix version in config.Defaults Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-01-05 11:44:49 -06:00
"github.com/nats-io/nats.go"
"github.com/sirupsen/logrus"
)
// RoomserverInternalAPI is an implementation of api.RoomserverInternalAPI
type RoomserverInternalAPI struct {
*input.Inputer
*query.Queryer
*perform.Inviter
*perform.Joiner
*perform.Peeker
Peeking over federation via MSC2444 (#1391) * a very very WIP first cut of peeking via MSC2753. doesn't yet compile or work. needs to actually add the peeking block into the sync response. checking in now before it gets any bigger, and to gather any initial feedback on the vague shape of it. * make PeekingDeviceSet private * add server_name param * blind stab at adding a `peek` section to /sync * make it build * make it launch * add peeking to getResponseWithPDUsForCompleteSync * cancel any peeks when we join a room * spell out how to runoutside of docker if you want speed * fix SQL * remove unnecessary txn for SelectPeeks * fix s/join/peek/ cargocult fail * HACK: Track goroutine IDs to determine when we write by the wrong thread To use: set `DENDRITE_TRACE_SQL=1` then grep for `unsafe` * Track partition offsets and only log unsafe for non-selects * Put redactions in the writer goroutine * Update filters on writer goroutine * wrap peek storage in goid hack * use exclusive writer, and MarkPeeksAsOld more efficiently * don't log ascii in binary at sql trace... * strip out empty roomd deltas * re-add txn to SelectPeeks * re-add accidentally deleted field * reject peeks for non-worldreadable rooms * move perform_peek * fix package * correctly refactor perform_peek * WIP of implementing MSC2444 * typo * Revert "Merge branch 'kegan/HACK-goid-sqlite-db-is-locked' into matthew/peeking" This reverts commit 3cebd8dbfbccdf82b7930b7b6eda92095ca6ef41, reversing changes made to ed4b3a58a7855acc43530693cc855b439edf9c7c. * (almost) make it build * clean up bad merge * support SendEventWithState with optional event * fix build & lint * fix build & lint * reinstate federated peeks in the roomserver (doh) * fix sql thinko * todo for authenticating state returned by /peek * support returning current state from QueryStateAndAuthChain * handle SS /peek * reimplement SS /peek to prod the RS to tell the FS about the peek * rename RemotePeeks as OutboundPeeks * rename remote_peeks_table as outbound_peeks_table * add perform_handle_remote_peek.go * flesh out federation doc * add inbound peeks table and hook it up * rename ambiguous RemotePeek as InboundPeek * rename FSAPI's PerformPeek as PerformOutboundPeek * setup inbound peeks db correctly * fix api.SendEventWithState with no event * track latestevent on /peek * go fmt * document the peek send stream race better * fix SendEventWithRewrite not to bail if handed a non-state event * add fixme * switch SS /peek to use SendEventWithRewrite * fix comment * use reverse topo ordering to find latest extrem * support postgres for federated peeking * go fmt * back out bogus go.mod change * Fix performOutboundPeekUsingServer * Fix getAuthChain -> GetAuthChain * Fix build issues * Fix build again * Fix getAuthChain -> GetAuthChain * Don't repeat outbound peeks for the same room ID to the same servers * Fix lint * Don't omitempty to appease sytest Co-authored-by: Kegan Dougal <kegan@matrix.org> Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2021-01-22 08:55:08 -06:00
*perform.InboundPeeker
*perform.Unpeeker
*perform.Leaver
*perform.Publisher
*perform.Backfiller
*perform.Forgetter
*perform.Upgrader
*perform.Admin
ProcessContext *process.ProcessContext
DB storage.Database
Cfg *config.RoomServer
Cache caching.RoomServerCaches
ServerName gomatrixserverlib.ServerName
KeyRing gomatrixserverlib.JSONVerifier
Roomserver/federation input refactor (#2104) * Put federation client functions into their own file * Look for missing auth events in RS input * Remove retrieveMissingAuthEvents from federation API * Logging * Sorta transplanted the code over * Use event origin failing all else * Don't get stuck on mutexes: * Add verifier * Don't mark state events with zero snapshot NID as not existing * Check missing state if not an outlier before storing the event * Reject instead of soft-fail, don't copy roominfo so much * Use synchronous contexts, limit time to fetch missing events * Clean up some commented out bits * Simplify `/send` endpoint significantly * Submit async * Report errors on sending to RS input * Set max payload in NATS to 16MB * Tweak metrics * Add `workerForRoom` for tidiness * Try skipping unmarshalling errors for RespMissingEvents * Track missing prev events separately to avoid calculating state when not possible * Tweak logic around checking missing state * Care about state when checking missing prev events * Don't check missing state for create events * Try that again * Handle create events better * Send create room events as new * Use given event kind when sending auth/state events * Revert "Use given event kind when sending auth/state events" This reverts commit 089d64d271b5fca8c104e1554711187420dbebca. * Only search for missing prev events or state for new events * Tweaks * We only have missing prev if we don't supply state * Room version tweaks * Allow async inputs again * Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed * Set timeouts on roomserver input tasks (need to decide what timeout makes sense) * Use work queue policy, deliver all on restart * Reduce chance of duplicates being sent by NATS * Limit the number of servers we attempt to reduce backpressure * Some review comment fixes * Tidy up a couple things * Don't limit servers, randomise order using map * Some context refactoring * Update gmsl * Don't resend create events * Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't * Exclude our own servername * Try backing off servers * Make excluding self behaviour optional * Exclude self from g_m_e * Update sytest-whitelist * Update consumers for the roomserver output stream * Remember to send outliers for state returned from /gme * Make full HTTP tests less upsetti * Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist * Remove debugging test * Fix blacklist again, remove unnecessary duplicate context * Clearer contexts, don't use background in case there's something happening there * Don't queue up events more than once in memory * Correctly identify create events when checking for state * Fill in gaps again in /gme code * Remove `AuthEventIDs` from `InputRoomEvent` * Remove stray field Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 08:29:14 -06:00
ServerACLs *acls.ServerACLs
fsAPI fsAPI.RoomserverFederationAPI
asAPI asAPI.AppServiceQueryAPI
NATSClient *nats.Conn
Roomserver/federation input refactor (#2104) * Put federation client functions into their own file * Look for missing auth events in RS input * Remove retrieveMissingAuthEvents from federation API * Logging * Sorta transplanted the code over * Use event origin failing all else * Don't get stuck on mutexes: * Add verifier * Don't mark state events with zero snapshot NID as not existing * Check missing state if not an outlier before storing the event * Reject instead of soft-fail, don't copy roominfo so much * Use synchronous contexts, limit time to fetch missing events * Clean up some commented out bits * Simplify `/send` endpoint significantly * Submit async * Report errors on sending to RS input * Set max payload in NATS to 16MB * Tweak metrics * Add `workerForRoom` for tidiness * Try skipping unmarshalling errors for RespMissingEvents * Track missing prev events separately to avoid calculating state when not possible * Tweak logic around checking missing state * Care about state when checking missing prev events * Don't check missing state for create events * Try that again * Handle create events better * Send create room events as new * Use given event kind when sending auth/state events * Revert "Use given event kind when sending auth/state events" This reverts commit 089d64d271b5fca8c104e1554711187420dbebca. * Only search for missing prev events or state for new events * Tweaks * We only have missing prev if we don't supply state * Room version tweaks * Allow async inputs again * Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed * Set timeouts on roomserver input tasks (need to decide what timeout makes sense) * Use work queue policy, deliver all on restart * Reduce chance of duplicates being sent by NATS * Limit the number of servers we attempt to reduce backpressure * Some review comment fixes * Tidy up a couple things * Don't limit servers, randomise order using map * Some context refactoring * Update gmsl * Don't resend create events * Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't * Exclude our own servername * Try backing off servers * Make excluding self behaviour optional * Exclude self from g_m_e * Update sytest-whitelist * Update consumers for the roomserver output stream * Remember to send outliers for state returned from /gme * Make full HTTP tests less upsetti * Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist * Remove debugging test * Fix blacklist again, remove unnecessary duplicate context * Clearer contexts, don't use background in case there's something happening there * Don't queue up events more than once in memory * Correctly identify create events when checking for state * Fill in gaps again in /gme code * Remove `AuthEventIDs` from `InputRoomEvent` * Remove stray field Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 08:29:14 -06:00
JetStream nats.JetStreamContext
Durable string
Add NATS JetStream support (#1866) * Add NATS JetStream support Update shopify/sarama * Fix addresses * Don't change Addresses in Defaults * Update saramajetstream * Add missing error check Keep typing events for at least one minute * Use all configured NATS addresses * Update saramajetstream * Try setting up with NATS * Make sure NATS uses own persistent directory (TODO: make this configurable) * Update go.mod/go.sum * Jetstream package * Various other refactoring * Build fixes * Config tweaks, make random jetstream storage path for CI * Disable interest policies * Try to sane default on jetstream base path * Try to use in-memory for CI * Restore storage/retention * Update nats.go dependency * Adapt changes to config * Remove unneeded TopicFor * Dep update * Revert "Remove unneeded TopicFor" This reverts commit f5a4e4a339b6f94ec215778dca22204adaa893d1. * Revert changes made to streams * Fix build problems * Update nats-server * Update go.mod/go.sum * Roomserver input API queuing using NATS * Fix topic naming * Prometheus metrics * More refactoring to remove saramajetstream * Add missing topic * Don't try to populate map that doesn't exist * Roomserver output topic * Update go.mod/go.sum * Message acknowledgements * Ack tweaks * Try to resume transaction re-sends * Try to resume transaction re-sends * Update to matrix-org/gomatrixserverlib@91dadfb * Remove internal.PartitionStorer from components that don't consume keychanges * Try to reduce re-allocations a bit in resolveConflictsV2 * Tweak delivery options on RS input * Publish send-to-device messages into correct JetStream subject * Async and sync roomserver input * Update dendrite-config.yaml * Remove roomserver tests for now (they need rewriting) * Remove roomserver test again (was merged back in) * Update documentation * Docker updates * More Docker updates * Update Docker readme again * Fix lint issues * Send final event in `processEvent` synchronously (since this might stop Sytest from being so upset) * Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that * Go 1.16 instead of Go 1.13 for upgrade tests and Complement * Revert "Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that" This reverts commit 368675283fc44501f227639811bdb16dd5deef8c. * Don't report any errors on `/send` to see what fun that creates * Fix panics on closed channel sends * Enforce state key matches sender * Do the same for leave * Various tweaks to make tests happier Squashed commit of the following: commit 13f9028e7a63662759ce7c55504a9d2423058668 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 15:47:14 2022 +0000 Do the same for leave commit e6be7f05c349fafbdddfe818337a17a60c867be1 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 15:33:42 2022 +0000 Enforce state key matches sender commit 85ede6d64bf10ce9b91cdd6d80f87350ee55242f Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 14:07:04 2022 +0000 Fix panics on closed channel sends commit 9755494a98bed62450f8001d8128e40481d27e15 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 13:38:22 2022 +0000 Don't report any errors on `/send` to see what fun that creates commit 3bb4f87b5dd56882febb4db5621db484c8789b7c Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 13:00:26 2022 +0000 Revert "Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that" This reverts commit 368675283fc44501f227639811bdb16dd5deef8c. commit fe2673ed7be9559eaca134424e403a4faca100b0 Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 12:09:34 2022 +0000 Go 1.16 instead of Go 1.13 for upgrade tests and Complement commit 368675283fc44501f227639811bdb16dd5deef8c Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 11:51:45 2022 +0000 Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that commit b028dfc08577bcf52e6cb498026e15fa5d46d07c Author: Neil Alexander <neilalexander@users.noreply.github.com> Date: Tue Jan 4 10:29:08 2022 +0000 Send final event in `processEvent` synchronously (since this might stop Sytest from being so upset) * Merge in NATS Server v2.6.6 and nats.go v1.13 into the in-process connection fork * Add `jetstream.WithJetStreamMessage` to make ack/nak-ing less messy, use process context in consumers * Fix consumer component name in federation API * Add comment explaining where streams are defined * Tweaks to roomserver input with comments * Finish that sentence that I apparently forgot to finish in INSTALL.md * Bump version number of config to 2 * Add comments around asynchronous sends to roomserver in processEventWithMissingState * More useful error message when the config version does not match * Set version in generate-config * Fix version in config.Defaults Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-01-05 11:44:49 -06:00
InputRoomEventTopic string // JetStream topic for new input room events
OutputRoomEventTopic string // JetStream topic for new output room events
PerspectiveServerNames []gomatrixserverlib.ServerName
}
func NewRoomserverAPI(
processCtx *process.ProcessContext, cfg *config.RoomServer, roomserverDB storage.Database,
consumer nats.JetStreamContext, nc *nats.Conn,
inputRoomEventTopic, outputRoomEventTopic string,
caches caching.RoomServerCaches, perspectiveServerNames []gomatrixserverlib.ServerName,
) *RoomserverInternalAPI {
serverACLs := acls.NewServerACLs(roomserverDB)
a := &RoomserverInternalAPI{
ProcessContext: processCtx,
DB: roomserverDB,
Cfg: cfg,
Cache: caches,
ServerName: cfg.Matrix.ServerName,
PerspectiveServerNames: perspectiveServerNames,
Roomserver/federation input refactor (#2104) * Put federation client functions into their own file * Look for missing auth events in RS input * Remove retrieveMissingAuthEvents from federation API * Logging * Sorta transplanted the code over * Use event origin failing all else * Don't get stuck on mutexes: * Add verifier * Don't mark state events with zero snapshot NID as not existing * Check missing state if not an outlier before storing the event * Reject instead of soft-fail, don't copy roominfo so much * Use synchronous contexts, limit time to fetch missing events * Clean up some commented out bits * Simplify `/send` endpoint significantly * Submit async * Report errors on sending to RS input * Set max payload in NATS to 16MB * Tweak metrics * Add `workerForRoom` for tidiness * Try skipping unmarshalling errors for RespMissingEvents * Track missing prev events separately to avoid calculating state when not possible * Tweak logic around checking missing state * Care about state when checking missing prev events * Don't check missing state for create events * Try that again * Handle create events better * Send create room events as new * Use given event kind when sending auth/state events * Revert "Use given event kind when sending auth/state events" This reverts commit 089d64d271b5fca8c104e1554711187420dbebca. * Only search for missing prev events or state for new events * Tweaks * We only have missing prev if we don't supply state * Room version tweaks * Allow async inputs again * Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed * Set timeouts on roomserver input tasks (need to decide what timeout makes sense) * Use work queue policy, deliver all on restart * Reduce chance of duplicates being sent by NATS * Limit the number of servers we attempt to reduce backpressure * Some review comment fixes * Tidy up a couple things * Don't limit servers, randomise order using map * Some context refactoring * Update gmsl * Don't resend create events * Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't * Exclude our own servername * Try backing off servers * Make excluding self behaviour optional * Exclude self from g_m_e * Update sytest-whitelist * Update consumers for the roomserver output stream * Remember to send outliers for state returned from /gme * Make full HTTP tests less upsetti * Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist * Remove debugging test * Fix blacklist again, remove unnecessary duplicate context * Clearer contexts, don't use background in case there's something happening there * Don't queue up events more than once in memory * Correctly identify create events when checking for state * Fill in gaps again in /gme code * Remove `AuthEventIDs` from `InputRoomEvent` * Remove stray field Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 08:29:14 -06:00
InputRoomEventTopic: inputRoomEventTopic,
OutputRoomEventTopic: outputRoomEventTopic,
JetStream: consumer,
NATSClient: nc,
Roomserver/federation input refactor (#2104) * Put federation client functions into their own file * Look for missing auth events in RS input * Remove retrieveMissingAuthEvents from federation API * Logging * Sorta transplanted the code over * Use event origin failing all else * Don't get stuck on mutexes: * Add verifier * Don't mark state events with zero snapshot NID as not existing * Check missing state if not an outlier before storing the event * Reject instead of soft-fail, don't copy roominfo so much * Use synchronous contexts, limit time to fetch missing events * Clean up some commented out bits * Simplify `/send` endpoint significantly * Submit async * Report errors on sending to RS input * Set max payload in NATS to 16MB * Tweak metrics * Add `workerForRoom` for tidiness * Try skipping unmarshalling errors for RespMissingEvents * Track missing prev events separately to avoid calculating state when not possible * Tweak logic around checking missing state * Care about state when checking missing prev events * Don't check missing state for create events * Try that again * Handle create events better * Send create room events as new * Use given event kind when sending auth/state events * Revert "Use given event kind when sending auth/state events" This reverts commit 089d64d271b5fca8c104e1554711187420dbebca. * Only search for missing prev events or state for new events * Tweaks * We only have missing prev if we don't supply state * Room version tweaks * Allow async inputs again * Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed * Set timeouts on roomserver input tasks (need to decide what timeout makes sense) * Use work queue policy, deliver all on restart * Reduce chance of duplicates being sent by NATS * Limit the number of servers we attempt to reduce backpressure * Some review comment fixes * Tidy up a couple things * Don't limit servers, randomise order using map * Some context refactoring * Update gmsl * Don't resend create events * Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't * Exclude our own servername * Try backing off servers * Make excluding self behaviour optional * Exclude self from g_m_e * Update sytest-whitelist * Update consumers for the roomserver output stream * Remember to send outliers for state returned from /gme * Make full HTTP tests less upsetti * Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist * Remove debugging test * Fix blacklist again, remove unnecessary duplicate context * Clearer contexts, don't use background in case there's something happening there * Don't queue up events more than once in memory * Correctly identify create events when checking for state * Fill in gaps again in /gme code * Remove `AuthEventIDs` from `InputRoomEvent` * Remove stray field Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 08:29:14 -06:00
Durable: cfg.Matrix.JetStream.Durable("RoomserverInputConsumer"),
ServerACLs: serverACLs,
Queryer: &query.Queryer{
DB: roomserverDB,
Cache: caches,
ServerName: cfg.Matrix.ServerName,
ServerACLs: serverACLs,
},
// perform-er structs get initialised when we have a federation sender to use
}
return a
}
// SetFederationInputAPI passes in a federation input API reference so that we can
// avoid the chicken-and-egg problem of both the roomserver input API and the
// federation input API being interdependent.
func (r *RoomserverInternalAPI) SetFederationAPI(fsAPI fsAPI.RoomserverFederationAPI, keyRing *gomatrixserverlib.KeyRing) {
r.fsAPI = fsAPI
r.KeyRing = keyRing
Roomserver/federation input refactor (#2104) * Put federation client functions into their own file * Look for missing auth events in RS input * Remove retrieveMissingAuthEvents from federation API * Logging * Sorta transplanted the code over * Use event origin failing all else * Don't get stuck on mutexes: * Add verifier * Don't mark state events with zero snapshot NID as not existing * Check missing state if not an outlier before storing the event * Reject instead of soft-fail, don't copy roominfo so much * Use synchronous contexts, limit time to fetch missing events * Clean up some commented out bits * Simplify `/send` endpoint significantly * Submit async * Report errors on sending to RS input * Set max payload in NATS to 16MB * Tweak metrics * Add `workerForRoom` for tidiness * Try skipping unmarshalling errors for RespMissingEvents * Track missing prev events separately to avoid calculating state when not possible * Tweak logic around checking missing state * Care about state when checking missing prev events * Don't check missing state for create events * Try that again * Handle create events better * Send create room events as new * Use given event kind when sending auth/state events * Revert "Use given event kind when sending auth/state events" This reverts commit 089d64d271b5fca8c104e1554711187420dbebca. * Only search for missing prev events or state for new events * Tweaks * We only have missing prev if we don't supply state * Room version tweaks * Allow async inputs again * Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed * Set timeouts on roomserver input tasks (need to decide what timeout makes sense) * Use work queue policy, deliver all on restart * Reduce chance of duplicates being sent by NATS * Limit the number of servers we attempt to reduce backpressure * Some review comment fixes * Tidy up a couple things * Don't limit servers, randomise order using map * Some context refactoring * Update gmsl * Don't resend create events * Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't * Exclude our own servername * Try backing off servers * Make excluding self behaviour optional * Exclude self from g_m_e * Update sytest-whitelist * Update consumers for the roomserver output stream * Remember to send outliers for state returned from /gme * Make full HTTP tests less upsetti * Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist * Remove debugging test * Fix blacklist again, remove unnecessary duplicate context * Clearer contexts, don't use background in case there's something happening there * Don't queue up events more than once in memory * Correctly identify create events when checking for state * Fill in gaps again in /gme code * Remove `AuthEventIDs` from `InputRoomEvent` * Remove stray field Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 08:29:14 -06:00
r.Inputer = &input.Inputer{
Cfg: r.Cfg,
ProcessContext: r.ProcessContext,
Roomserver/federation input refactor (#2104) * Put federation client functions into their own file * Look for missing auth events in RS input * Remove retrieveMissingAuthEvents from federation API * Logging * Sorta transplanted the code over * Use event origin failing all else * Don't get stuck on mutexes: * Add verifier * Don't mark state events with zero snapshot NID as not existing * Check missing state if not an outlier before storing the event * Reject instead of soft-fail, don't copy roominfo so much * Use synchronous contexts, limit time to fetch missing events * Clean up some commented out bits * Simplify `/send` endpoint significantly * Submit async * Report errors on sending to RS input * Set max payload in NATS to 16MB * Tweak metrics * Add `workerForRoom` for tidiness * Try skipping unmarshalling errors for RespMissingEvents * Track missing prev events separately to avoid calculating state when not possible * Tweak logic around checking missing state * Care about state when checking missing prev events * Don't check missing state for create events * Try that again * Handle create events better * Send create room events as new * Use given event kind when sending auth/state events * Revert "Use given event kind when sending auth/state events" This reverts commit 089d64d271b5fca8c104e1554711187420dbebca. * Only search for missing prev events or state for new events * Tweaks * We only have missing prev if we don't supply state * Room version tweaks * Allow async inputs again * Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed * Set timeouts on roomserver input tasks (need to decide what timeout makes sense) * Use work queue policy, deliver all on restart * Reduce chance of duplicates being sent by NATS * Limit the number of servers we attempt to reduce backpressure * Some review comment fixes * Tidy up a couple things * Don't limit servers, randomise order using map * Some context refactoring * Update gmsl * Don't resend create events * Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't * Exclude our own servername * Try backing off servers * Make excluding self behaviour optional * Exclude self from g_m_e * Update sytest-whitelist * Update consumers for the roomserver output stream * Remember to send outliers for state returned from /gme * Make full HTTP tests less upsetti * Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist * Remove debugging test * Fix blacklist again, remove unnecessary duplicate context * Clearer contexts, don't use background in case there's something happening there * Don't queue up events more than once in memory * Correctly identify create events when checking for state * Fill in gaps again in /gme code * Remove `AuthEventIDs` from `InputRoomEvent` * Remove stray field Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 08:29:14 -06:00
DB: r.DB,
InputRoomEventTopic: r.InputRoomEventTopic,
OutputRoomEventTopic: r.OutputRoomEventTopic,
JetStream: r.JetStream,
NATSClient: r.NATSClient,
Durable: nats.Durable(r.Durable),
Roomserver/federation input refactor (#2104) * Put federation client functions into their own file * Look for missing auth events in RS input * Remove retrieveMissingAuthEvents from federation API * Logging * Sorta transplanted the code over * Use event origin failing all else * Don't get stuck on mutexes: * Add verifier * Don't mark state events with zero snapshot NID as not existing * Check missing state if not an outlier before storing the event * Reject instead of soft-fail, don't copy roominfo so much * Use synchronous contexts, limit time to fetch missing events * Clean up some commented out bits * Simplify `/send` endpoint significantly * Submit async * Report errors on sending to RS input * Set max payload in NATS to 16MB * Tweak metrics * Add `workerForRoom` for tidiness * Try skipping unmarshalling errors for RespMissingEvents * Track missing prev events separately to avoid calculating state when not possible * Tweak logic around checking missing state * Care about state when checking missing prev events * Don't check missing state for create events * Try that again * Handle create events better * Send create room events as new * Use given event kind when sending auth/state events * Revert "Use given event kind when sending auth/state events" This reverts commit 089d64d271b5fca8c104e1554711187420dbebca. * Only search for missing prev events or state for new events * Tweaks * We only have missing prev if we don't supply state * Room version tweaks * Allow async inputs again * Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed * Set timeouts on roomserver input tasks (need to decide what timeout makes sense) * Use work queue policy, deliver all on restart * Reduce chance of duplicates being sent by NATS * Limit the number of servers we attempt to reduce backpressure * Some review comment fixes * Tidy up a couple things * Don't limit servers, randomise order using map * Some context refactoring * Update gmsl * Don't resend create events * Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't * Exclude our own servername * Try backing off servers * Make excluding self behaviour optional * Exclude self from g_m_e * Update sytest-whitelist * Update consumers for the roomserver output stream * Remember to send outliers for state returned from /gme * Make full HTTP tests less upsetti * Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist * Remove debugging test * Fix blacklist again, remove unnecessary duplicate context * Clearer contexts, don't use background in case there's something happening there * Don't queue up events more than once in memory * Correctly identify create events when checking for state * Fill in gaps again in /gme code * Remove `AuthEventIDs` from `InputRoomEvent` * Remove stray field Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 08:29:14 -06:00
ServerName: r.Cfg.Matrix.ServerName,
FSAPI: fsAPI,
KeyRing: keyRing,
ACLs: r.ServerACLs,
Queryer: r.Queryer,
}
r.Inviter = &perform.Inviter{
DB: r.DB,
Cfg: r.Cfg,
FSAPI: r.fsAPI,
Inputer: r.Inputer,
}
r.Joiner = &perform.Joiner{
ServerName: r.Cfg.Matrix.ServerName,
Cfg: r.Cfg,
DB: r.DB,
FSAPI: r.fsAPI,
RSAPI: r,
Inputer: r.Inputer,
Queryer: r.Queryer,
}
r.Peeker = &perform.Peeker{
ServerName: r.Cfg.Matrix.ServerName,
Cfg: r.Cfg,
DB: r.DB,
FSAPI: r.fsAPI,
Inputer: r.Inputer,
}
Peeking over federation via MSC2444 (#1391) * a very very WIP first cut of peeking via MSC2753. doesn't yet compile or work. needs to actually add the peeking block into the sync response. checking in now before it gets any bigger, and to gather any initial feedback on the vague shape of it. * make PeekingDeviceSet private * add server_name param * blind stab at adding a `peek` section to /sync * make it build * make it launch * add peeking to getResponseWithPDUsForCompleteSync * cancel any peeks when we join a room * spell out how to runoutside of docker if you want speed * fix SQL * remove unnecessary txn for SelectPeeks * fix s/join/peek/ cargocult fail * HACK: Track goroutine IDs to determine when we write by the wrong thread To use: set `DENDRITE_TRACE_SQL=1` then grep for `unsafe` * Track partition offsets and only log unsafe for non-selects * Put redactions in the writer goroutine * Update filters on writer goroutine * wrap peek storage in goid hack * use exclusive writer, and MarkPeeksAsOld more efficiently * don't log ascii in binary at sql trace... * strip out empty roomd deltas * re-add txn to SelectPeeks * re-add accidentally deleted field * reject peeks for non-worldreadable rooms * move perform_peek * fix package * correctly refactor perform_peek * WIP of implementing MSC2444 * typo * Revert "Merge branch 'kegan/HACK-goid-sqlite-db-is-locked' into matthew/peeking" This reverts commit 3cebd8dbfbccdf82b7930b7b6eda92095ca6ef41, reversing changes made to ed4b3a58a7855acc43530693cc855b439edf9c7c. * (almost) make it build * clean up bad merge * support SendEventWithState with optional event * fix build & lint * fix build & lint * reinstate federated peeks in the roomserver (doh) * fix sql thinko * todo for authenticating state returned by /peek * support returning current state from QueryStateAndAuthChain * handle SS /peek * reimplement SS /peek to prod the RS to tell the FS about the peek * rename RemotePeeks as OutboundPeeks * rename remote_peeks_table as outbound_peeks_table * add perform_handle_remote_peek.go * flesh out federation doc * add inbound peeks table and hook it up * rename ambiguous RemotePeek as InboundPeek * rename FSAPI's PerformPeek as PerformOutboundPeek * setup inbound peeks db correctly * fix api.SendEventWithState with no event * track latestevent on /peek * go fmt * document the peek send stream race better * fix SendEventWithRewrite not to bail if handed a non-state event * add fixme * switch SS /peek to use SendEventWithRewrite * fix comment * use reverse topo ordering to find latest extrem * support postgres for federated peeking * go fmt * back out bogus go.mod change * Fix performOutboundPeekUsingServer * Fix getAuthChain -> GetAuthChain * Fix build issues * Fix build again * Fix getAuthChain -> GetAuthChain * Don't repeat outbound peeks for the same room ID to the same servers * Fix lint * Don't omitempty to appease sytest Co-authored-by: Kegan Dougal <kegan@matrix.org> Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2021-01-22 08:55:08 -06:00
r.InboundPeeker = &perform.InboundPeeker{
DB: r.DB,
Inputer: r.Inputer,
}
r.Unpeeker = &perform.Unpeeker{
ServerName: r.Cfg.Matrix.ServerName,
Cfg: r.Cfg,
DB: r.DB,
FSAPI: r.fsAPI,
Inputer: r.Inputer,
}
r.Leaver = &perform.Leaver{
Cfg: r.Cfg,
DB: r.DB,
FSAPI: r.fsAPI,
Inputer: r.Inputer,
}
r.Publisher = &perform.Publisher{
DB: r.DB,
}
r.Backfiller = &perform.Backfiller{
ServerName: r.ServerName,
DB: r.DB,
FSAPI: r.fsAPI,
KeyRing: r.KeyRing,
// Perspective servers are trusted to not lie about server keys, so we will also
// prefer these servers when backfilling (assuming they are in the room) rather
// than trying random servers
PreferServers: r.PerspectiveServerNames,
}
r.Forgetter = &perform.Forgetter{
DB: r.DB,
}
r.Upgrader = &perform.Upgrader{
Cfg: r.Cfg,
URSAPI: r,
}
r.Admin = &perform.Admin{
DB: r.DB,
Cfg: r.Cfg,
Inputer: r.Inputer,
Queryer: r.Queryer,
}
if err := r.Inputer.Start(); err != nil {
logrus.WithError(err).Panic("failed to start roomserver input API")
}
}
func (r *RoomserverInternalAPI) SetUserAPI(userAPI userapi.UserInternalAPI) {
r.Leaver.UserAPI = userAPI
}
func (r *RoomserverInternalAPI) SetAppserviceAPI(asAPI asAPI.AppServiceQueryAPI) {
r.asAPI = asAPI
}
func (r *RoomserverInternalAPI) PerformInvite(
ctx context.Context,
req *api.PerformInviteRequest,
res *api.PerformInviteResponse,
) error {
outputEvents, err := r.Inviter.PerformInvite(ctx, req, res)
if err != nil {
sentry.CaptureException(err)
return err
}
if len(outputEvents) == 0 {
return nil
}
return r.WriteOutputEvents(req.Event.RoomID(), outputEvents)
}
func (r *RoomserverInternalAPI) PerformLeave(
ctx context.Context,
req *api.PerformLeaveRequest,
res *api.PerformLeaveResponse,
) error {
outputEvents, err := r.Leaver.PerformLeave(ctx, req, res)
if err != nil {
sentry.CaptureException(err)
return err
}
if len(outputEvents) == 0 {
return nil
}
return r.WriteOutputEvents(req.RoomID, outputEvents)
}
func (r *RoomserverInternalAPI) PerformForget(
ctx context.Context,
req *api.PerformForgetRequest,
resp *api.PerformForgetResponse,
) error {
return r.Forgetter.PerformForget(ctx, req, resp)
}