syncthing/cmd/stdiscosrv/amqp.go

262 lines
5.7 KiB
Go
Raw Permalink Normal View History

2024-06-03 05:13:21 +00:00
// Copyright (C) 2024 The Syncthing Authors.
//
// This Source Code Form is subject to the terms of the Mozilla Public
// License, v. 2.0. If a copy of the MPL was not distributed with this file,
// You can obtain one at https://mozilla.org/MPL/2.0/.
package main
import (
"context"
"fmt"
"io"
"log"
2024-06-03 05:13:21 +00:00
amqp "github.com/rabbitmq/amqp091-go"
"github.com/thejerf/suture/v4"
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
"google.golang.org/protobuf/proto"
"github.com/syncthing/syncthing/internal/gen/discosrv"
"github.com/syncthing/syncthing/internal/protoutil"
"github.com/syncthing/syncthing/lib/protocol"
2024-06-03 05:13:21 +00:00
)
type amqpReplicator struct {
suture.Service
broker string
sender *amqpSender
receiver *amqpReceiver
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
outbox chan *discosrv.ReplicationRecord
2024-06-03 05:13:21 +00:00
}
func newAMQPReplicator(broker, clientID string, db database) *amqpReplicator {
svc := suture.New("amqpReplicator", suture.Spec{PassThroughPanics: true})
sender := &amqpSender{
broker: broker,
clientID: clientID,
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
outbox: make(chan *discosrv.ReplicationRecord, replicationOutboxSize),
2024-06-03 05:13:21 +00:00
}
svc.Add(sender)
receiver := &amqpReceiver{
broker: broker,
clientID: clientID,
db: db,
}
svc.Add(receiver)
return &amqpReplicator{
Service: svc,
broker: broker,
sender: sender,
receiver: receiver,
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
outbox: make(chan *discosrv.ReplicationRecord, replicationOutboxSize),
2024-06-03 05:13:21 +00:00
}
}
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
func (s *amqpReplicator) send(key *protocol.DeviceID, ps []*discosrv.DatabaseAddress, seen int64) {
2024-06-03 05:13:21 +00:00
s.sender.send(key, ps, seen)
}
type amqpSender struct {
broker string
clientID string
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
outbox chan *discosrv.ReplicationRecord
2024-06-03 05:13:21 +00:00
}
func (s *amqpSender) Serve(ctx context.Context) error {
conn, ch, err := amqpChannel(s.broker)
if err != nil {
return err
}
defer ch.Close()
defer conn.Close()
buf := make([]byte, 1024)
for {
select {
case rec := <-s.outbox:
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
size := proto.Size(rec)
2024-06-03 05:13:21 +00:00
if len(buf) < size {
buf = make([]byte, size)
}
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
n, err := protoutil.MarshalTo(buf, rec)
2024-06-03 05:13:21 +00:00
if err != nil {
replicationSendsTotal.WithLabelValues("error").Inc()
return fmt.Errorf("replication marshal: %w", err)
}
err = ch.PublishWithContext(ctx,
"discovery", // exchange
"", // routing key
false, // mandatory
false, // immediate
amqp.Publishing{
ContentType: "application/protobuf",
Body: buf[:n],
AppId: s.clientID,
})
if err != nil {
replicationSendsTotal.WithLabelValues("error").Inc()
return fmt.Errorf("replication publish: %w", err)
}
replicationSendsTotal.WithLabelValues("success").Inc()
case <-ctx.Done():
return nil
}
}
}
func (s *amqpSender) String() string {
return fmt.Sprintf("amqpSender(%q)", s.broker)
}
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
func (s *amqpSender) send(key *protocol.DeviceID, ps []*discosrv.DatabaseAddress, seen int64) {
item := &discosrv.ReplicationRecord{
Key: key[:],
2024-06-03 05:13:21 +00:00
Addresses: ps,
Seen: seen,
}
// The send should never block. The inbox is suitably buffered for at
// least a few seconds of stalls, which shouldn't happen in practice.
select {
case s.outbox <- item:
default:
replicationSendsTotal.WithLabelValues("drop").Inc()
}
}
type amqpReceiver struct {
broker string
clientID string
db database
}
func (s *amqpReceiver) Serve(ctx context.Context) error {
conn, ch, err := amqpChannel(s.broker)
if err != nil {
return err
}
defer ch.Close()
defer conn.Close()
msgs, err := amqpConsume(ch)
if err != nil {
return err
}
for {
select {
case msg, ok := <-msgs:
if !ok {
return fmt.Errorf("subscription closed: %w", io.EOF)
}
// ignore messages from ourself
if msg.AppId == s.clientID {
continue
}
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
var rec discosrv.ReplicationRecord
if err := proto.Unmarshal(msg.Body, &rec); err != nil {
2024-06-03 05:13:21 +00:00
replicationRecvsTotal.WithLabelValues("error").Inc()
return fmt.Errorf("replication unmarshal: %w", err)
}
id, err := protocol.DeviceIDFromBytes(rec.Key)
if err != nil {
id, err = protocol.DeviceIDFromString(string(rec.Key))
}
if err != nil {
log.Println("Replication device ID:", err)
replicationRecvsTotal.WithLabelValues("error").Inc()
continue
}
2024-06-03 05:13:21 +00:00
if err := s.db.merge(&id, rec.Addresses, rec.Seen); err != nil {
2024-06-03 05:13:21 +00:00
return fmt.Errorf("replication database merge: %w", err)
}
replicationRecvsTotal.WithLabelValues("success").Inc()
case <-ctx.Done():
return nil
}
}
}
func (s *amqpReceiver) String() string {
return fmt.Sprintf("amqpReceiver(%q)", s.broker)
}
func amqpChannel(dst string) (*amqp.Connection, *amqp.Channel, error) {
conn, err := amqp.Dial(dst)
if err != nil {
return nil, nil, fmt.Errorf("AMQP dial: %w", err)
}
ch, err := conn.Channel()
if err != nil {
return nil, nil, fmt.Errorf("AMQP channel: %w", err)
}
err = ch.ExchangeDeclare(
"discovery", // name
"fanout", // type
false, // durable
false, // auto-deleted
false, // internal
false, // no-wait
nil, // arguments
)
if err != nil {
return nil, nil, fmt.Errorf("AMQP declare exchange: %w", err)
}
return conn, ch, nil
}
func amqpConsume(ch *amqp.Channel) (<-chan amqp.Delivery, error) {
q, err := ch.QueueDeclare(
"", // name
false, // durable
false, // delete when unused
true, // exclusive
false, // no-wait
nil, // arguments
)
if err != nil {
return nil, fmt.Errorf("AMQP declare queue: %w", err)
}
err = ch.QueueBind(
q.Name, // queue name
"", // routing key
"discovery", // exchange
false,
nil,
)
if err != nil {
return nil, fmt.Errorf("AMQP bind queue: %w", err)
}
msgs, err := ch.Consume(
q.Name, // queue
"", // consumer
true, // auto-ack
false, // exclusive
false, // no-local
false, // no-wait
nil, // args
)
if err != nil {
return nil, fmt.Errorf("AMQP consume: %w", err)
}
return msgs, nil
}