ipfs-cluster/test/ipfs_mock.go

509 lines
10 KiB
Go
Raw Normal View History

package test
import (
"context"
"encoding/json"
"fmt"
"io/ioutil"
"net/http"
"net/http/httptest"
"net/url"
"strconv"
"strings"
"sync"
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
"testing"
"time"
"github.com/ipfs/ipfs-cluster/api"
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
"github.com/ipfs/ipfs-cluster/datastore/inmem"
"github.com/ipfs/ipfs-cluster/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
"github.com/ipfs/ipfs-cluster/state/dsstate"
"github.com/multiformats/go-multihash"
cid "github.com/ipfs/go-cid"
cors "github.com/rs/cors"
)
// Some values used by the ipfs mock
const (
Fix #382 (again): A better strategy for handling proxy headers This changes the current strategy to extract headers from the IPFS daemon to use them for hijacked endpoints in the proxy. The ipfs daemon is a bit of a mess and what we were doing is not really reliable, specially when it comes to setting CORS headers right (which we were not doing). The new approach is: * For every hijacked request, make an OPTIONS request to the same path, with the given Origin, to the IPFS daemon and extract some CORS headers from that. Use those in the hijacked response * Avoid hijacking OPTIONS request, they should always go through so the IPFS daemon controls all the CORS-preflight things as it wants. * Similar to before, have a only-once-triggered request to extract other interesting or custom headers from a fixed IPFS endpoint. This allows us to have the proxy forward other custom headers and to catch `Access-Control-Expose-Methods`. The difference is that the endpoint use for this and the additional headers are configurable by the user (but with hidden configuration options because this is quite exotic from regular usage). Now the implementation: * Replaced the standard Muxer with gorilla/mux (I have also taken the change to update the gxed version to the latest tag). This gives us much better matching control over routes and allows us to not handle OPTIONS requests. * This allows also to remove the extractArgument code and have proper handlers for the endpoints passing command arguments as the last segment of the URL. A very simple handler that wraps the default ones can be used to extract the argument from the url and put it in the query. Overall much cleaner this way. * No longer capture interesting headers from any random proxied request. This made things complicated with a wrapping handler. We will just trigger the one request to do it when we need it. * When preparing the headers for the hijacked responses: * Trigger the OPTIONS request and figure out which CORS things we should set * Set the additional headers (perhaps triggering a POST request to fetch them) * Set our own headers. * Moved all the headers stuff to a new headers.go file. * Added configuration options (hidden by default) to: * Customize the extract headers endpoint * Customize what additional headers are extracted * Use HTTPs when talking to the IPFS API * I haven't tested this, but I did not want to have hardcoded 'http://' urls around, as before. * Added extra testing for this, and tested manually a lot comparing the daemon original output with our hijacked endpoint outputs while looking at the API traffic with ngrep and making sure the requets happen as expected. Also tested with IPFS companion in FF and Chrome. License: MIT Signed-off-by: Hector Sanjuan <code@hector.link>
2019-01-10 19:03:59 +00:00
IpfsCustomHeaderName = "X-Custom-Header"
IpfsTimeHeaderName = "X-Time-Now"
Fix #382 (again): A better strategy for handling proxy headers This changes the current strategy to extract headers from the IPFS daemon to use them for hijacked endpoints in the proxy. The ipfs daemon is a bit of a mess and what we were doing is not really reliable, specially when it comes to setting CORS headers right (which we were not doing). The new approach is: * For every hijacked request, make an OPTIONS request to the same path, with the given Origin, to the IPFS daemon and extract some CORS headers from that. Use those in the hijacked response * Avoid hijacking OPTIONS request, they should always go through so the IPFS daemon controls all the CORS-preflight things as it wants. * Similar to before, have a only-once-triggered request to extract other interesting or custom headers from a fixed IPFS endpoint. This allows us to have the proxy forward other custom headers and to catch `Access-Control-Expose-Methods`. The difference is that the endpoint use for this and the additional headers are configurable by the user (but with hidden configuration options because this is quite exotic from regular usage). Now the implementation: * Replaced the standard Muxer with gorilla/mux (I have also taken the change to update the gxed version to the latest tag). This gives us much better matching control over routes and allows us to not handle OPTIONS requests. * This allows also to remove the extractArgument code and have proper handlers for the endpoints passing command arguments as the last segment of the URL. A very simple handler that wraps the default ones can be used to extract the argument from the url and put it in the query. Overall much cleaner this way. * No longer capture interesting headers from any random proxied request. This made things complicated with a wrapping handler. We will just trigger the one request to do it when we need it. * When preparing the headers for the hijacked responses: * Trigger the OPTIONS request and figure out which CORS things we should set * Set the additional headers (perhaps triggering a POST request to fetch them) * Set our own headers. * Moved all the headers stuff to a new headers.go file. * Added configuration options (hidden by default) to: * Customize the extract headers endpoint * Customize what additional headers are extracted * Use HTTPs when talking to the IPFS API * I haven't tested this, but I did not want to have hardcoded 'http://' urls around, as before. * Added extra testing for this, and tested manually a lot comparing the daemon original output with our hijacked endpoint outputs while looking at the API traffic with ngrep and making sure the requets happen as expected. Also tested with IPFS companion in FF and Chrome. License: MIT Signed-off-by: Hector Sanjuan <code@hector.link>
2019-01-10 19:03:59 +00:00
IpfsCustomHeaderValue = "42"
IpfsACAOrigin = "myorigin"
IpfsErrFromNotPinned = "'from' cid was not recursively pinned already"
Fix #382 (again): A better strategy for handling proxy headers This changes the current strategy to extract headers from the IPFS daemon to use them for hijacked endpoints in the proxy. The ipfs daemon is a bit of a mess and what we were doing is not really reliable, specially when it comes to setting CORS headers right (which we were not doing). The new approach is: * For every hijacked request, make an OPTIONS request to the same path, with the given Origin, to the IPFS daemon and extract some CORS headers from that. Use those in the hijacked response * Avoid hijacking OPTIONS request, they should always go through so the IPFS daemon controls all the CORS-preflight things as it wants. * Similar to before, have a only-once-triggered request to extract other interesting or custom headers from a fixed IPFS endpoint. This allows us to have the proxy forward other custom headers and to catch `Access-Control-Expose-Methods`. The difference is that the endpoint use for this and the additional headers are configurable by the user (but with hidden configuration options because this is quite exotic from regular usage). Now the implementation: * Replaced the standard Muxer with gorilla/mux (I have also taken the change to update the gxed version to the latest tag). This gives us much better matching control over routes and allows us to not handle OPTIONS requests. * This allows also to remove the extractArgument code and have proper handlers for the endpoints passing command arguments as the last segment of the URL. A very simple handler that wraps the default ones can be used to extract the argument from the url and put it in the query. Overall much cleaner this way. * No longer capture interesting headers from any random proxied request. This made things complicated with a wrapping handler. We will just trigger the one request to do it when we need it. * When preparing the headers for the hijacked responses: * Trigger the OPTIONS request and figure out which CORS things we should set * Set the additional headers (perhaps triggering a POST request to fetch them) * Set our own headers. * Moved all the headers stuff to a new headers.go file. * Added configuration options (hidden by default) to: * Customize the extract headers endpoint * Customize what additional headers are extracted * Use HTTPs when talking to the IPFS API * I haven't tested this, but I did not want to have hardcoded 'http://' urls around, as before. * Added extra testing for this, and tested manually a lot comparing the daemon original output with our hijacked endpoint outputs while looking at the API traffic with ngrep and making sure the requets happen as expected. Also tested with IPFS companion in FF and Chrome. License: MIT Signed-off-by: Hector Sanjuan <code@hector.link>
2019-01-10 19:03:59 +00:00
)
// IpfsMock is an ipfs daemon mock which should sustain the functionality used by ipfscluster.
type IpfsMock struct {
server *httptest.Server
Addr string
Port int
pinMap state.State
BlockStore map[string][]byte
reqCounts map[string]int
reqCounter chan string
closeMux sync.Mutex
closed bool
}
type mockPinResp struct {
Pins []string
Progress int `json:",omitempty"`
}
type mockPinType struct {
Type string
}
type mockPinLsResp struct {
Keys map[string]mockPinType
}
type ipfsErr struct {
Code int
Message string
}
type mockIDResp struct {
ID string
Addresses []string
}
type mockRepoStatResp struct {
RepoSize uint64
NumObjects uint64
StorageMax uint64
}
type mockConfigResp struct {
Datastore struct {
StorageMax string
}
}
type mockRefsResp struct {
Ref string
Err string
}
type mockSwarmPeersResp struct {
Peers []mockIpfsPeer
}
type mockIpfsPeer struct {
Peer string
}
type mockBlockPutResp struct {
Key string
}
type mockRepoGCResp struct {
Key cid.Cid `json:",omitempty"`
Error string `json:",omitempty"`
}
// NewIpfsMock returns a new mock.
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 NewIpfsMock(t *testing.T) *IpfsMock {
store := inmem.New()
st, err := dsstate.New(store, "", dsstate.DefaultHandle())
if err != nil {
t.Fatal(err)
}
m := &IpfsMock{
pinMap: st,
BlockStore: make(map[string][]byte),
reqCounts: make(map[string]int),
reqCounter: make(chan string, 100),
}
Fix #382 (again): A better strategy for handling proxy headers This changes the current strategy to extract headers from the IPFS daemon to use them for hijacked endpoints in the proxy. The ipfs daemon is a bit of a mess and what we were doing is not really reliable, specially when it comes to setting CORS headers right (which we were not doing). The new approach is: * For every hijacked request, make an OPTIONS request to the same path, with the given Origin, to the IPFS daemon and extract some CORS headers from that. Use those in the hijacked response * Avoid hijacking OPTIONS request, they should always go through so the IPFS daemon controls all the CORS-preflight things as it wants. * Similar to before, have a only-once-triggered request to extract other interesting or custom headers from a fixed IPFS endpoint. This allows us to have the proxy forward other custom headers and to catch `Access-Control-Expose-Methods`. The difference is that the endpoint use for this and the additional headers are configurable by the user (but with hidden configuration options because this is quite exotic from regular usage). Now the implementation: * Replaced the standard Muxer with gorilla/mux (I have also taken the change to update the gxed version to the latest tag). This gives us much better matching control over routes and allows us to not handle OPTIONS requests. * This allows also to remove the extractArgument code and have proper handlers for the endpoints passing command arguments as the last segment of the URL. A very simple handler that wraps the default ones can be used to extract the argument from the url and put it in the query. Overall much cleaner this way. * No longer capture interesting headers from any random proxied request. This made things complicated with a wrapping handler. We will just trigger the one request to do it when we need it. * When preparing the headers for the hijacked responses: * Trigger the OPTIONS request and figure out which CORS things we should set * Set the additional headers (perhaps triggering a POST request to fetch them) * Set our own headers. * Moved all the headers stuff to a new headers.go file. * Added configuration options (hidden by default) to: * Customize the extract headers endpoint * Customize what additional headers are extracted * Use HTTPs when talking to the IPFS API * I haven't tested this, but I did not want to have hardcoded 'http://' urls around, as before. * Added extra testing for this, and tested manually a lot comparing the daemon original output with our hijacked endpoint outputs while looking at the API traffic with ngrep and making sure the requets happen as expected. Also tested with IPFS companion in FF and Chrome. License: MIT Signed-off-by: Hector Sanjuan <code@hector.link>
2019-01-10 19:03:59 +00:00
go m.countRequests()
Fix #382 (again): A better strategy for handling proxy headers This changes the current strategy to extract headers from the IPFS daemon to use them for hijacked endpoints in the proxy. The ipfs daemon is a bit of a mess and what we were doing is not really reliable, specially when it comes to setting CORS headers right (which we were not doing). The new approach is: * For every hijacked request, make an OPTIONS request to the same path, with the given Origin, to the IPFS daemon and extract some CORS headers from that. Use those in the hijacked response * Avoid hijacking OPTIONS request, they should always go through so the IPFS daemon controls all the CORS-preflight things as it wants. * Similar to before, have a only-once-triggered request to extract other interesting or custom headers from a fixed IPFS endpoint. This allows us to have the proxy forward other custom headers and to catch `Access-Control-Expose-Methods`. The difference is that the endpoint use for this and the additional headers are configurable by the user (but with hidden configuration options because this is quite exotic from regular usage). Now the implementation: * Replaced the standard Muxer with gorilla/mux (I have also taken the change to update the gxed version to the latest tag). This gives us much better matching control over routes and allows us to not handle OPTIONS requests. * This allows also to remove the extractArgument code and have proper handlers for the endpoints passing command arguments as the last segment of the URL. A very simple handler that wraps the default ones can be used to extract the argument from the url and put it in the query. Overall much cleaner this way. * No longer capture interesting headers from any random proxied request. This made things complicated with a wrapping handler. We will just trigger the one request to do it when we need it. * When preparing the headers for the hijacked responses: * Trigger the OPTIONS request and figure out which CORS things we should set * Set the additional headers (perhaps triggering a POST request to fetch them) * Set our own headers. * Moved all the headers stuff to a new headers.go file. * Added configuration options (hidden by default) to: * Customize the extract headers endpoint * Customize what additional headers are extracted * Use HTTPs when talking to the IPFS API * I haven't tested this, but I did not want to have hardcoded 'http://' urls around, as before. * Added extra testing for this, and tested manually a lot comparing the daemon original output with our hijacked endpoint outputs while looking at the API traffic with ngrep and making sure the requets happen as expected. Also tested with IPFS companion in FF and Chrome. License: MIT Signed-off-by: Hector Sanjuan <code@hector.link>
2019-01-10 19:03:59 +00:00
mux := http.NewServeMux()
mux.HandleFunc("/", m.handler)
c := cors.New(cors.Options{
AllowedOrigins: []string{IpfsACAOrigin},
AllowedMethods: []string{"POST"},
ExposedHeaders: []string{"X-Stream-Output", "X-Chunked-Output", "X-Content-Length"},
AllowCredentials: true, // because IPFS does it, even if for no reason.
})
corsHandler := c.Handler(mux)
ts := httptest.NewServer(corsHandler)
m.server = ts
url, _ := url.Parse(ts.URL)
h := strings.Split(url.Host, ":")
i, _ := strconv.Atoi(h[1])
m.Port = i
m.Addr = h[0]
return m
}
func (m *IpfsMock) countRequests() {
for str := range m.reqCounter {
m.reqCounts[str]++
}
}
// GetCount allows to get the number of times and endpoint was called.
// Do not use concurrently to requests happening.
func (m *IpfsMock) GetCount(path string) int {
return m.reqCounts[path]
}
// FIXME: what if IPFS API changes?
func (m *IpfsMock) handler(w http.ResponseWriter, r *http.Request) {
ctx := context.Background()
p := r.URL.Path
Fix #382 (again): A better strategy for handling proxy headers This changes the current strategy to extract headers from the IPFS daemon to use them for hijacked endpoints in the proxy. The ipfs daemon is a bit of a mess and what we were doing is not really reliable, specially when it comes to setting CORS headers right (which we were not doing). The new approach is: * For every hijacked request, make an OPTIONS request to the same path, with the given Origin, to the IPFS daemon and extract some CORS headers from that. Use those in the hijacked response * Avoid hijacking OPTIONS request, they should always go through so the IPFS daemon controls all the CORS-preflight things as it wants. * Similar to before, have a only-once-triggered request to extract other interesting or custom headers from a fixed IPFS endpoint. This allows us to have the proxy forward other custom headers and to catch `Access-Control-Expose-Methods`. The difference is that the endpoint use for this and the additional headers are configurable by the user (but with hidden configuration options because this is quite exotic from regular usage). Now the implementation: * Replaced the standard Muxer with gorilla/mux (I have also taken the change to update the gxed version to the latest tag). This gives us much better matching control over routes and allows us to not handle OPTIONS requests. * This allows also to remove the extractArgument code and have proper handlers for the endpoints passing command arguments as the last segment of the URL. A very simple handler that wraps the default ones can be used to extract the argument from the url and put it in the query. Overall much cleaner this way. * No longer capture interesting headers from any random proxied request. This made things complicated with a wrapping handler. We will just trigger the one request to do it when we need it. * When preparing the headers for the hijacked responses: * Trigger the OPTIONS request and figure out which CORS things we should set * Set the additional headers (perhaps triggering a POST request to fetch them) * Set our own headers. * Moved all the headers stuff to a new headers.go file. * Added configuration options (hidden by default) to: * Customize the extract headers endpoint * Customize what additional headers are extracted * Use HTTPs when talking to the IPFS API * I haven't tested this, but I did not want to have hardcoded 'http://' urls around, as before. * Added extra testing for this, and tested manually a lot comparing the daemon original output with our hijacked endpoint outputs while looking at the API traffic with ngrep and making sure the requets happen as expected. Also tested with IPFS companion in FF and Chrome. License: MIT Signed-off-by: Hector Sanjuan <code@hector.link>
2019-01-10 19:03:59 +00:00
w.Header().Set(IpfsCustomHeaderName, IpfsCustomHeaderValue)
w.Header().Set("Server", "ipfs-mock")
w.Header().Set(IpfsTimeHeaderName, fmt.Sprintf("%d", time.Now().Unix()))
endp := strings.TrimPrefix(p, "/api/v0/")
m.reqCounter <- endp
switch endp {
case "id":
resp := mockIDResp{
ID: PeerID1.Pretty(),
Addresses: []string{
"/ip4/0.0.0.0/tcp/1234",
"/ip6/::/tcp/1234",
},
}
j, _ := json.Marshal(resp)
w.Write(j)
case "pin/add":
arg, ok := extractCid(r.URL)
if !ok {
goto ERROR
}
if arg == ErrorCid.String() {
goto ERROR
}
c, err := cid.Decode(arg)
if err != nil {
goto ERROR
}
mode := extractMode(r.URL)
opts := api.PinOptions{
Mode: mode,
}
pinObj := api.PinWithOpts(c, opts)
m.pinMap.Add(ctx, pinObj)
resp := mockPinResp{
Pins: []string{arg},
}
if c.Equals(SlowCid1) {
for i := 0; i <= 10; i++ {
time.Sleep(1 * time.Second)
resp.Progress = i
j, _ := json.Marshal(resp)
w.Write(j)
}
} else {
j, _ := json.Marshal(resp)
w.Write(j)
}
case "pin/rm":
arg, ok := extractCid(r.URL)
if !ok {
goto ERROR
}
c, err := cid.Decode(arg)
if err != nil {
goto ERROR
}
m.pinMap.Rm(ctx, c)
resp := mockPinResp{
Pins: []string{arg},
}
j, _ := json.Marshal(resp)
w.Write(j)
case "pin/update":
args := r.URL.Query()["arg"]
if len(args) != 2 {
goto ERROR
}
fromStr := args[0]
toStr := args[1]
from, err := cid.Decode(fromStr)
if err != nil {
goto ERROR
}
to, err := cid.Decode(toStr)
if err != nil {
goto ERROR
}
pin, err := m.pinMap.Get(ctx, from)
if err != nil {
w.WriteHeader(http.StatusInternalServerError)
resp := ipfsErr{0, IpfsErrFromNotPinned}
j, _ := json.Marshal(resp)
w.Write(j)
return
}
pin.Cid = to
err = m.pinMap.Add(ctx, pin)
if err != nil {
goto ERROR
}
resp := mockPinResp{
Pins: []string{from.String(), to.String()},
}
j, _ := json.Marshal(resp)
w.Write(j)
case "pin/ls":
arg, ok := extractCid(r.URL)
if !ok {
rMap := make(map[string]mockPinType)
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
pins, err := m.pinMap.List(ctx)
if err != nil {
goto ERROR
}
for _, p := range pins {
rMap[p.Cid.String()] = mockPinType{p.Mode.String()}
}
j, _ := json.Marshal(mockPinLsResp{rMap})
w.Write(j)
break
}
cidStr := arg
c, err := cid.Decode(cidStr)
if err != nil {
goto ERROR
}
pinObj, err := m.pinMap.Get(ctx, c)
if err != nil && err != state.ErrNotFound {
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
goto ERROR
}
if err == state.ErrNotFound {
w.WriteHeader(http.StatusInternalServerError)
resp := ipfsErr{0, fmt.Sprintf("Path '%s' is not pinned", cidStr)}
j, _ := json.Marshal(resp)
w.Write(j)
return
}
if c.Equals(Cid4) {
// this a v1 cid. Do not return default-base32 but base58btc encoding of it
w.Write([]byte(`{ "Keys": { "zCT5htkdztJi3x4zBNHo8TRvGHPLTdHUdCLKgTGMgQcRKSLoWxK1": { "Type": "recursive" }}}`))
return
}
rMap := make(map[string]mockPinType)
rMap[cidStr] = mockPinType{pinObj.Mode.String()}
j, _ := json.Marshal(mockPinLsResp{rMap})
w.Write(j)
case "swarm/connect":
arg, ok := extractCid(r.URL)
if !ok {
goto ERROR
}
addr := arg
splits := strings.Split(addr, "/")
pid := splits[len(splits)-1]
resp := struct {
Strings []string
}{
Strings: []string{fmt.Sprintf("connect %s success", pid)},
}
j, _ := json.Marshal(resp)
w.Write(j)
case "swarm/peers":
peer1 := mockIpfsPeer{
Peer: PeerID4.Pretty(),
}
peer2 := mockIpfsPeer{
Peer: PeerID5.Pretty(),
}
resp := mockSwarmPeersResp{
Peers: []mockIpfsPeer{peer1, peer2},
}
j, _ := json.Marshal(resp)
w.Write(j)
case "block/put":
// Get the data and retun the hash
mpr, err := r.MultipartReader()
if err != nil {
goto ERROR
}
part, err := mpr.NextPart()
if err != nil {
goto ERROR
}
data, err := ioutil.ReadAll(part)
if err != nil {
goto ERROR
}
// Parse cid from data and format and add to mock block-store
query := r.URL.Query()
format := cid.Codecs[query.Get("format")]
mhType := multihash.Names[query.Get("mhtype")]
mhLen, _ := strconv.Atoi(query.Get("mhLen"))
var builder cid.Builder
if format == cid.DagProtobuf && mhType == multihash.SHA2_256 {
builder = cid.V0Builder{}
} else {
builder = cid.V1Builder{
Codec: format,
MhType: mhType,
MhLength: mhLen,
}
}
c, err := builder.Sum(data)
if err != nil {
goto ERROR
}
m.BlockStore[c.String()] = data
resp := mockBlockPutResp{
Key: c.String(),
}
j, _ := json.Marshal(resp)
w.Write(j)
case "block/get":
query := r.URL.Query()
arg, ok := query["arg"]
if !ok {
goto ERROR
}
if len(arg) != 1 {
goto ERROR
}
data, ok := m.BlockStore[arg[0]]
if !ok {
goto ERROR
}
w.Write(data)
case "repo/gc":
// It assumes `/repo/gc` with parameter `stream-errors=true`
enc := json.NewEncoder(w)
resp := []mockRepoGCResp{
{
Key: Cid1,
},
{
Key: Cid2,
},
{
Key: Cid3,
},
{
Key: Cid4,
},
{
Error: "no link by that name",
},
}
for _, r := range resp {
if err := enc.Encode(&r); err != nil {
goto ERROR
}
}
case "repo/stat":
sizeOnly := r.URL.Query().Get("size-only")
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
list, err := m.pinMap.List(ctx)
if err != nil {
goto ERROR
}
len := len(list)
numObjs := uint64(len)
if sizeOnly == "true" {
numObjs = 0
}
resp := mockRepoStatResp{
RepoSize: uint64(len) * 1000,
NumObjects: numObjs,
StorageMax: 10000000000, //10 GB
}
j, _ := json.Marshal(resp)
w.Write(j)
Add PinPath/UnpinPath support. Squashed commit of the following: commit 38cf569c6aed77c46ee4e0f8baa4d1a9daf8f03e Merge: d125f69 aaada42 Author: Hector Sanjuan <hsanjuan@users.noreply.github.com> Date: Wed Feb 20 11:02:00 2019 +0000 Merge pull request #634 from ipfs/issue_450 Support PinPath, UnpinPath (resolve before pinning) commit aaada42054e1f1c7b2abb1270859d0de41a0e5d8 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Feb 19 22:16:25 2019 +0530 formatResponse accepts api.Pin and not api.PinSerial commit b5da4bea045865814cc422da71827b44ddd44b90 Merge: ba59036 cc8dd7e Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Feb 19 21:36:46 2019 +0530 Merge branch 'master' into issue_450 commit ba5903649c1df1dba20f4d6f7e3573d6fe24921f Merge: f002914 d59880c Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Mon Feb 18 08:41:11 2019 +0530 Merge branch 'issue_450' of github.com:ipfs/ipfs-cluster into issue_450 commit f00291494c0c02621c2296cbb7ac71e4c23aa9ec Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Mon Feb 18 08:31:39 2019 +0530 PinPath: more improvements Added tracing for new methods commit d59880c338eaa8214fe06b4f930a540793d78407 Merge: 0ca4c7c b4f0eb3 Author: Hector Sanjuan <hsanjuan@users.noreply.github.com> Date: Wed Feb 13 15:22:49 2019 +0000 Merge branch 'master' into issue_450 commit 0ca4c7c3b0670ed9c8279f8274d36e3485c10030 Merge: d35017a ecef9ea Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Feb 12 13:10:13 2019 +0530 Merge branch 'master' into issue_450 commit d35017a8de91ca9fc9a9a047c48c75134cee9f98 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Feb 12 13:07:25 2019 +0530 PinPath: more improvements - Worth having `PinOptions` as a separate field in the struct and constructing the query in the test with ToQuery() - sharness: "intialization" line can be placed outside the tests at the top commit 68e3b90417ffbad89d41a70ac81d85f9037f8848 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Sun Feb 10 21:43:50 2019 +0530 Using if-continue pattern instead of if-else License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 3c29799f3b85be328b27508332ab92049d8b82f3 Merge: 956790b 4324889 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Thu Feb 7 10:25:52 2019 +0530 Merge branch 'master' into issue_450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 956790b381db9858e4194f983e898b07dc51ba66 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Feb 6 21:11:20 2019 +0530 Removing resolved path License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 7191cc46cedfbec116a9746937e28881b50ca044 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Feb 6 16:45:07 2019 +0530 Fix go vet License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit f8b3d5b63b1b7569e2a3e0d82894fd4491c246c4 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Feb 6 16:07:03 2019 +0530 Fixed linting error License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 23c57eb467755a1f21387a1615a7f34e97348053 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Feb 6 09:20:41 2019 +0530 Fixed tests License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 0caedd94aefeb3b6649dedc214cb4b849ace2ea4 Merge: 17e555e 5a7ee1d Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Feb 6 00:07:10 2019 +0530 Merge branch 'master' into issue_450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 17e555e4a7c574413df90aac70c5cc29cab98f54 Author: Hector Sanjuan <code@hector.link> Date: Tue Feb 5 16:58:50 2019 +0000 PinPath: address some feedback + improvements * Changed client's Pin() API and PinPath to be consistent * Added helper methods to turn PinPath to query and back * Make code and tests build * Use TestCidResolved everywhere * Fix cluster.PinPath arguments * Fix formatting of responses with --no-status * Make tests readable and call Fatal when needed * Use a pathTestCases variable commit f0e7369c47c5ddadc8ed45df5fd2d4d9b2d42b38 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Feb 5 18:34:26 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) Addressed review comments as in https://github.com/ipfs/ipfs-cluster/pull/634#pullrequestreview-198751932 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit a8b4f181d2d7afed32ee41331dfaab19fd66a173 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Jan 29 22:41:27 2019 +0530 Fixing tests License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit e39b95ca19e4d75506f4f492678245ef13936a44 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Jan 29 14:52:53 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) - PinPath and UnpinPath should return api.Pin - PinPath should accept pin options - Removing duplicate logic for Resolve from cluster - And many other review comments https://github.com/ipfs/ipfs-cluster/pull/634#pullrequestreview-195509504 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit d146075126320896665ba58d337a13789f68ea86 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Jan 23 17:08:41 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) PinPath(in both rest and rpc) should return a serializable struct in the form `{"\":"Q...cid..string..."}` (as used in "github.com/ipfs/go-cid" to marshal and unmarshal) License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 1f4869568a8adb450275257154ea3a26d03a30f3 Merge: 7acfd28 a244af9 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Jan 23 07:18:56 2019 +0530 Merge branch 'master' into issue_450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 7acfd282732ddf2282a67d4f9d0170a494eb3ed4 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Jan 22 18:14:32 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) - RPC must always use serializable structs - In command, just use pin with path as cid is also a valid path - Addressing many other small review comments as in https://github.com/ipfs/ipfs-cluster/pull/634#pullrequestreview-192122534 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 36905041e1e3f0b204942030aab3ab7b5b9e4d62 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Jan 16 09:36:42 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) Extra logic for path checking should go into resolve so that it can be properly reused Added sharness tests License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 9116bda3534e77bb391d873051bb520a1b01a326 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Jan 16 08:08:07 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) error strings should not be capitalized Fixes #450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit ca7e61861374f456300a85ddc0374e594f74f963 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Jan 15 23:40:25 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) Tests Fixes #450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 522fbcd899f01c01680375561a32a87464157c0a Merge: f1a56ab f7bc468 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Jan 15 10:40:54 2019 +0530 Merge branch 'master' into issue_450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit f1a56ab925fb74c0c44273a4524afa4843cf757f Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Mon Jan 14 20:58:17 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) - IPFS Connector should act as a pure IPFS client, any extra logic should go to cluster.go - Use cid.Undef, instead of cid.Cid{} Fixes #450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit c83b91054f6774f1f9d4930cfc3f1fa28236f57c Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Thu Jan 10 08:57:17 2019 +0530 Support PinPath, UnpinPath(resolve before pinning) - Separate handlers, methods and rpc apis for PinPath and UnpinPath from Pin and Unpin - Support ipld paths as well Fixes #450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 719dff88129366ce3ccb5e04cb6f8082a0915c5c Merge: 91ceb47 21170c4 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Jan 9 19:38:35 2019 +0530 Merge branch 'issue_450_old' into HEAD License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 91ceb4796259ca7ef2974ec43e6a278a12796b13 Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Wed Jan 9 19:36:41 2019 +0530 Revert "WIP: Figure out why test does not impleme" This reverts commit 28a3a3f25dce6f296c8cbef86221644c099a7e75. License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> commit 28a3a3f25dce6f296c8cbef86221644c099a7e75 Author: cd10012 <ced361@nyu.edu> Date: Tue Jul 24 23:23:10 2018 -0400 WIP: Figure out why test does not implement IPFSConnector interface... License: MIT Signed-off-by: cd10012 <ced361@nyu.edu> commit 21170c48e77e69583db64544b08120a9baf40d8d Author: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Date: Tue Jan 8 10:37:59 2019 +0530 Support PinPath, UnpinPath (resolve before pinning) This commit adds API support for pinning using path `POST /pins/<ipfs or ipns path>` and `DELETE /pins/<ipfs or ipns path>` will resolve the path into a cid and perform perform pinning or unpinning Fixes #450 License: MIT Signed-off-by: Kishan Mohanbhai Sagathiya <kishansagathiya@gmail.com> Co-authored-by: Hector Sanjuan <hector@protocol.ai> License: MIT Signed-off-by: Hector Sanjuan <hector@protocol.ai>
2019-02-20 11:07:50 +00:00
case "resolve":
w.Write([]byte("{\"Path\":\"" + "/ipfs/" + CidResolved.String() + "\"}"))
case "config/show":
resp := mockConfigResp{
Datastore: struct {
StorageMax string
}{
StorageMax: "10G",
},
}
j, _ := json.Marshal(resp)
w.Write(j)
case "refs":
arg, ok := extractCid(r.URL)
if !ok {
goto ERROR
}
resp := mockRefsResp{
Ref: arg,
}
j, _ := json.Marshal(resp)
if arg == SlowCid1.String() {
for i := 0; i <= 5; i++ {
time.Sleep(2 * time.Second)
w.Write(j)
}
} else {
w.Write(j)
}
case "version":
w.Write([]byte("{\"Version\":\"m.o.c.k\"}"))
default:
w.WriteHeader(http.StatusNotFound)
}
return
ERROR:
w.WriteHeader(http.StatusInternalServerError)
}
// Close closes the mock server. It's important to call after each test or
// the listeners are left hanging around.
func (m *IpfsMock) Close() {
m.closeMux.Lock()
defer m.closeMux.Unlock()
if !m.closed {
m.closed = true
m.server.Close()
close(m.reqCounter)
}
}
// extractCidAndMode extracts the cid argument from a url.URL, either via
// the query string parameters or from the url path itself.
func extractCid(u *url.URL) (string, bool) {
arg := u.Query().Get("arg")
if arg != "" {
return arg, true
}
p := strings.TrimPrefix(u.Path, "/api/v0/")
segs := strings.Split(p, "/")
if len(segs) > 2 {
return segs[len(segs)-1], true
}
return "", false
}
func extractMode(u *url.URL) api.PinMode {
return api.PinModeFromString(u.Query().Get("type"))
}