ipfs-cluster/consensus/crdt/consensus_test.go

497 lines
9.9 KiB
Go
Raw Normal View History

Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
package crdt
import (
"context"
"fmt"
"testing"
"time"
"github.com/ipfs-cluster/ipfs-cluster/api"
"github.com/ipfs-cluster/ipfs-cluster/datastore/inmem"
"github.com/ipfs-cluster/ipfs-cluster/test"
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
2023-03-27 19:49:08 +00:00
ipns "github.com/ipfs/boxo/ipns"
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
libp2p "github.com/libp2p/go-libp2p"
dht "github.com/libp2p/go-libp2p-kad-dht"
2020-04-14 20:03:24 +00:00
dual "github.com/libp2p/go-libp2p-kad-dht/dual"
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
pubsub "github.com/libp2p/go-libp2p-pubsub"
record "github.com/libp2p/go-libp2p-record"
2023-03-27 19:49:08 +00:00
host "github.com/libp2p/go-libp2p/core/host"
peerstore "github.com/libp2p/go-libp2p/core/peerstore"
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
routedhost "github.com/libp2p/go-libp2p/p2p/host/routed"
)
2020-04-14 20:03:24 +00:00
func makeTestingHost(t *testing.T) (host.Host, *pubsub.PubSub, *dual.DHT) {
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
ctx := context.Background()
h, err := libp2p.New(
libp2p.ListenAddrStrings("/ip4/127.0.0.1/tcp/0"),
)
if err != nil {
t.Fatal(err)
}
psub, err := pubsub.NewGossipSub(
ctx,
h,
pubsub.WithMessageSigning(true),
pubsub.WithStrictSignatureVerification(true),
)
if err != nil {
h.Close()
t.Fatal(err)
}
2020-04-14 20:03:24 +00:00
idht, err := dual.New(ctx, h,
2020-09-02 10:06:47 +00:00
dual.DHTOption(dht.NamespacedValidator("pk", record.PublicKeyValidator{})),
dual.DHTOption(dht.NamespacedValidator("ipns", ipns.Validator{KeyBook: h.Peerstore()})),
dual.DHTOption(dht.Concurrency(10)),
dual.DHTOption(dht.RoutingTableRefreshPeriod(200*time.Millisecond)),
dual.DHTOption(dht.RoutingTableRefreshQueryTimeout(100*time.Millisecond)),
)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if err != nil {
h.Close()
t.Fatal(err)
}
rHost := routedhost.Wrap(h, idht)
return rHost, psub, idht
}
func testingConsensus(t *testing.T, idn int) *Consensus {
cfg := &Config{}
cfg.Default()
2021-04-30 17:45:01 +00:00
return testingConsensusWithCfg(t, idn, cfg)
}
func testingConsensusWithCfg(t *testing.T, idn int, cfg *Config) *Consensus {
h, psub, dht := makeTestingHost(t)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
cfg.DatastoreNamespace = fmt.Sprintf("crdttest-%d", idn)
cfg.hostShutdown = true
cc, err := New(h, dht, psub, cfg, inmem.New())
if err != nil {
t.Fatal("cannot create Consensus:", err)
}
cc.SetClient(test.NewMockRPCClientWithHost(t, h))
<-cc.Ready(context.Background())
return cc
}
func clean(t *testing.T, cc *Consensus) {
err := cc.Clean(context.Background())
if err != nil {
t.Error(err)
}
}
func testPin(c api.Cid) api.Pin {
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
p := api.PinCid(c)
p.ReplicationFactorMin = -1
p.ReplicationFactorMax = -1
return p
}
func TestShutdownConsensus(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
defer clean(t, cc)
err := cc.Shutdown(ctx)
if err != nil {
t.Fatal("Consensus cannot shutdown:", err)
}
err = cc.Shutdown(ctx) // should be fine to shutdown twice
if err != nil {
t.Fatal("Consensus should be able to shutdown several times")
}
}
func TestConsensusPin(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
defer clean(t, cc)
defer cc.Shutdown(ctx)
err := cc.LogPin(ctx, testPin(test.Cid1))
if err != nil {
t.Error(err)
}
time.Sleep(250 * time.Millisecond)
st, err := cc.State(ctx)
if err != nil {
t.Fatal("error getting state:", err)
}
out := make(chan api.Pin, 10)
err = st.List(ctx, out)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if err != nil {
t.Fatal(err)
}
var pins []api.Pin
for p := range out {
pins = append(pins, p)
}
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if len(pins) != 1 || !pins[0].Cid.Equals(test.Cid1) {
t.Error("the added pin should be in the state")
}
}
func TestConsensusUnpin(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
defer clean(t, cc)
defer cc.Shutdown(ctx)
err := cc.LogPin(ctx, testPin(test.Cid1))
if err != nil {
t.Error(err)
}
err = cc.LogUnpin(ctx, api.PinCid(test.Cid1))
if err != nil {
t.Error(err)
}
}
func TestConsensusUpdate(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
defer clean(t, cc)
defer cc.Shutdown(ctx)
// Pin first
pin := testPin(test.Cid1)
pin.Type = api.ShardType
err := cc.LogPin(ctx, pin)
if err != nil {
t.Fatal(err)
}
time.Sleep(500 * time.Millisecond)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
// Update pin
pin.Reference = &test.Cid2
err = cc.LogPin(ctx, pin)
if err != nil {
t.Error(err)
}
time.Sleep(500 * time.Millisecond)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
st, err := cc.State(ctx)
if err != nil {
t.Fatal("error getting state:", err)
}
// Channel will not block sending because plenty of space
out := make(chan api.Pin, 100)
err = st.List(ctx, out)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if err != nil {
t.Fatal(err)
}
var pins []api.Pin
for p := range out {
pins = append(pins, p)
}
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if len(pins) != 1 || !pins[0].Cid.Equals(test.Cid1) {
t.Fatal("the added pin should be in the state")
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
}
if !pins[0].Reference.Equals(test.Cid2) {
t.Error("pin updated incorrectly")
}
}
func TestConsensusAddRmPeer(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
cc2 := testingConsensus(t, 2)
defer clean(t, cc)
defer clean(t, cc)
defer cc.Shutdown(ctx)
defer cc2.Shutdown(ctx)
cc.host.Peerstore().AddAddrs(cc2.host.ID(), cc2.host.Addrs(), peerstore.PermanentAddrTTL)
_, err := cc.host.Network().DialPeer(ctx, cc2.host.ID())
if err != nil {
t.Fatal(err)
}
time.Sleep(100 * time.Millisecond)
err = cc.AddPeer(ctx, cc2.host.ID())
if err != nil {
t.Error("could not add peer:", err)
}
err = cc2.Trust(ctx, cc.host.ID())
if err != nil {
t.Error("could not trust peer:", err)
}
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
// Make a pin on peer1 and check it arrived to peer2
err = cc.LogPin(ctx, testPin(test.Cid1))
if err != nil {
t.Error(err)
}
time.Sleep(500 * time.Millisecond)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
st, err := cc2.State(ctx)
if err != nil {
t.Fatal("error getting state:", err)
}
out := make(chan api.Pin, 100)
err = st.List(ctx, out)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if err != nil {
t.Fatal(err)
}
var pins []api.Pin
for p := range out {
pins = append(pins, p)
}
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if len(pins) != 1 || !pins[0].Cid.Equals(test.Cid1) {
t.Error("the added pin should be in the state")
}
err = cc2.RmPeer(ctx, cc.host.ID())
if err == nil {
t.Error("crdt consensus should not remove pins")
}
}
2019-05-13 21:11:02 +00:00
func TestConsensusDistrustPeer(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
cc2 := testingConsensus(t, 2)
defer clean(t, cc)
defer clean(t, cc)
defer cc.Shutdown(ctx)
defer cc2.Shutdown(ctx)
cc.host.Peerstore().AddAddrs(cc2.host.ID(), cc2.host.Addrs(), peerstore.PermanentAddrTTL)
_, err := cc.host.Network().DialPeer(ctx, cc2.host.ID())
if err != nil {
t.Fatal(err)
}
time.Sleep(100 * time.Millisecond)
err = cc2.Trust(ctx, cc.host.ID())
if err != nil {
t.Error("could not trust peer:", err)
}
// Make a pin on peer1 and check it arrived to peer2
err = cc.LogPin(ctx, testPin(test.Cid1))
if err != nil {
t.Error(err)
}
time.Sleep(250 * time.Millisecond)
err = cc2.Distrust(ctx, cc.host.ID())
if err != nil {
t.Error("could not distrust peer:", err)
}
// Another pin should never get to peer2
err = cc.LogPin(ctx, testPin(test.Cid2))
if err != nil {
t.Error(err)
}
// Verify we only got the first pin
st, err := cc2.State(ctx)
if err != nil {
t.Fatal("error getting state:", err)
}
out := make(chan api.Pin, 10)
err = st.List(ctx, out)
2019-05-13 21:11:02 +00:00
if err != nil {
t.Fatal(err)
}
var pins []api.Pin
for p := range out {
pins = append(pins, p)
}
2019-05-13 21:11:02 +00:00
if len(pins) != 1 || !pins[0].Cid.Equals(test.Cid1) {
t.Error("only first pin should be in the state")
2019-05-13 21:11:02 +00:00
}
}
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
func TestPeers(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
defer clean(t, cc)
defer cc.Shutdown(ctx)
peers, err := cc.Peers(ctx)
if err != nil {
t.Fatal(err)
}
// 1 is ourselves and the other comes from rpc
// mock PeerMonitorLatestMetrics
if len(peers) != 2 {
t.Error("unexpected number of peers")
}
}
func TestOfflineState(t *testing.T) {
ctx := context.Background()
cc := testingConsensus(t, 1)
defer clean(t, cc)
defer cc.Shutdown(ctx)
// Make pin 1
err := cc.LogPin(ctx, testPin(test.Cid1))
if err != nil {
t.Error(err)
}
// Make pin 2
err = cc.LogPin(ctx, testPin(test.Cid2))
if err != nil {
t.Error(err)
}
err = cc.Shutdown(ctx)
if err != nil {
t.Fatal(err)
}
offlineState, err := OfflineState(cc.config, cc.store)
if err != nil {
t.Fatal(err)
}
out := make(chan api.Pin, 100)
err = offlineState.List(ctx, out)
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if err != nil {
t.Fatal(err)
}
var pins []api.Pin
for p := range out {
pins = append(pins, p)
}
Consensus: add new "crdt" consensus component This adds a new "crdt" consensus component using go-ds-crdt. This implies several refactors to fully make cluster consensus-component independent: * Delete mapstate and fully adopt dsstate (after people have migrated). * Return errors from state methods rather than ignoring them. * Add a new "datastore" modules so that we can configure datastores in the main configuration like other components. * Let the consensus components fully define the "state.State". Thus, they do not receive the state, they receive the storage where we put the state (a go-datastore). * Allow to customize how the monitor component obtains Peers() (the current peerset), including avoiding using the current peerset. At the moment the crdt consensus uses the monitoring component to define the current peerset. Therefore the monitor component cannot rely on the consensus component to produce a peerset. * Re-factor/re-implementation of "ipfs-cluster-service state" operations. Includes the dissapearance of the "migrate" one. The CRDT consensus component defines creates a crdt-datastore (with ipfs-lite) and uses it to intitialize a dssate. Thus the crdt-store is elegantly wrapped. Any modifications to the state get automatically replicated to other peers. We store all the CRDT DAG blocks in the local datastore. The consensus components only expose a ReadOnly state, as any modifications to the shared state should happen through them. DHT and PubSub facilities must now be created outside of Cluster and passed in so they can be re-used by different components.
2019-02-20 14:24:25 +00:00
if len(pins) != 2 {
t.Error("there should be two pins in the state")
}
}
2021-04-30 17:45:01 +00:00
func TestBatching(t *testing.T) {
ctx := context.Background()
cfg := &Config{}
cfg.Default()
cfg.Batching.MaxBatchSize = 3
cfg.Batching.MaxBatchAge = 1 * time.Second
cc := testingConsensusWithCfg(t, 1, cfg)
defer clean(t, cc)
defer cc.Shutdown(ctx)
st, err := cc.State(ctx)
if err != nil {
t.Fatal("error getting state:", err)
}
// Pin something
err = cc.LogPin(ctx, testPin(test.Cid1))
if err != nil {
t.Error(err)
}
time.Sleep(250 * time.Millisecond)
out := make(chan api.Pin, 100)
err = st.List(ctx, out)
2021-04-30 17:45:01 +00:00
if err != nil {
t.Fatal(err)
}
var pins []api.Pin
for p := range out {
pins = append(pins, p)
}
2021-04-30 17:45:01 +00:00
if len(pins) != 0 {
t.Error("pin should not be pinned yet as it is being batched")
}
// Trigger batch auto-commit by time
time.Sleep(time.Second)
out = make(chan api.Pin, 100)
err = st.List(ctx, out)
2021-04-30 17:45:01 +00:00
if err != nil {
t.Fatal(err)
}
pins = nil
for p := range out {
pins = append(pins, p)
}
2021-04-30 17:45:01 +00:00
if len(pins) != 1 || !pins[0].Cid.Equals(test.Cid1) {
t.Error("the added pin should be in the state")
}
// Pin 4 things, and check that 3 are committed
for _, c := range []api.Cid{test.Cid2, test.Cid3, test.Cid4, test.Cid5} {
2021-04-30 17:45:01 +00:00
err = cc.LogPin(ctx, testPin(c))
if err != nil {
t.Error(err)
}
}
// Give a chance for things to persist
time.Sleep(250 * time.Millisecond)
out = make(chan api.Pin, 100)
err = st.List(ctx, out)
2021-04-30 17:45:01 +00:00
if err != nil {
t.Fatal(err)
}
pins = nil
for p := range out {
pins = append(pins, p)
}
2021-04-30 17:45:01 +00:00
if len(pins) != 4 {
t.Error("expected 4 items pinned")
}
// wait for the last pin
time.Sleep(time.Second)
out = make(chan api.Pin, 100)
err = st.List(ctx, out)
2021-04-30 17:45:01 +00:00
if err != nil {
t.Fatal(err)
}
pins = nil
for p := range out {
pins = append(pins, p)
}
2021-04-30 17:45:01 +00:00
if len(pins) != 5 {
t.Error("expected 5 items pinned")
}
}