storj/satellite/contact/service.go
JT Olio da9ca0c650 testplanet/satellite: reduce the number of places default values need to be configured
Satellites set their configuration values to default values using
cfgstruct, however, it turns out our tests don't test these values
at all! Instead, they have a completely separate definition system
that is easy to forget about.

As is to be expected, these values have drifted, and it appears
in a few cases test planet is testing unreasonable values that we
won't see in production, or perhaps worse, features enabled in
production were missed and weren't enabled in testplanet.

This change makes it so all values are configured the same,
systematic way, so it's easy to see when test values are different
than dev values or release values, and it's less hard to forget
to enable features in testplanet.

In terms of reviewing, this change should be actually fairly
easy to review, considering private/testplanet/satellite.go keeps
the current config system and the new one and confirms that they
result in identical configurations, so you can be certain that
nothing was missed and the config is all correct.
You can also check the config lock to see what actual config
values changed.

Change-Id: I6715d0794887f577e21742afcf56fd2b9d12170e
2021-06-01 22:14:17 +00:00

151 lines
4.9 KiB
Go

// Copyright (C) 2019 Storj Labs, Inc.
// See LICENSE for copying information.
package contact
import (
"context"
"fmt"
"sync"
"time"
"github.com/zeebo/errs"
"go.uber.org/zap"
"storj.io/common/pb"
"storj.io/common/rpc"
"storj.io/common/rpc/rpcstatus"
"storj.io/common/storj"
"storj.io/storj/private/quic"
"storj.io/storj/satellite/overlay"
)
// Config contains configurable values for contact service.
type Config struct {
ExternalAddress string `user:"true" help:"the public address of the node, useful for nodes behind NAT" default:""`
Timeout time.Duration `help:"timeout for pinging storage nodes" default:"10m0s" testDefault:"1m"`
RateLimitInterval time.Duration `help:"the amount of time that should happen between contact attempts usually" releaseDefault:"10m0s" devDefault:"1ns"`
RateLimitBurst int `help:"the maximum burst size for the contact rate limit token bucket" releaseDefault:"2" devDefault:"1000"`
RateLimitCacheSize int `help:"the number of nodes or addresses to keep token buckets for" default:"1000"`
}
// Service is the contact service between storage nodes and satellites.
// It is responsible for updating general node information like address and capacity.
// It is also responsible for updating peer identity information for verifying signatures from that node.
//
// architecture: Service
type Service struct {
log *zap.Logger
mutex sync.Mutex
self *overlay.NodeDossier
overlay *overlay.Service
peerIDs overlay.PeerIdentities
dialer rpc.Dialer
timeout time.Duration
idLimiter *RateLimiter
}
// NewService creates a new contact service.
func NewService(log *zap.Logger, self *overlay.NodeDossier, overlay *overlay.Service, peerIDs overlay.PeerIdentities, dialer rpc.Dialer, config Config) *Service {
return &Service{
log: log,
self: self,
overlay: overlay,
peerIDs: peerIDs,
dialer: dialer,
timeout: config.Timeout,
idLimiter: NewRateLimiter(config.RateLimitInterval, config.RateLimitBurst, config.RateLimitCacheSize),
}
}
// Local returns the satellite node dossier.
func (service *Service) Local() overlay.NodeDossier {
service.mutex.Lock()
defer service.mutex.Unlock()
return *service.self
}
// Close closes resources.
func (service *Service) Close() error { return nil }
// PingBack pings the node to test connectivity.
func (service *Service) PingBack(ctx context.Context, nodeurl storj.NodeURL) (_ bool, _ bool, _ string, err error) {
defer mon.Task()(&ctx)(&err)
if service.timeout > 0 {
var cancel func()
ctx, cancel = context.WithTimeout(ctx, service.timeout)
defer cancel()
}
pingNodeSuccess := true
var pingErrorMessage string
var pingNodeSuccessQUIC bool
client, err := dialNodeURL(ctx, service.dialer, nodeurl)
if err != nil {
// If there is an error from trying to dial and ping the node, return that error as
// pingErrorMessage and not as the err. We want to use this info to update
// node contact info and do not want to terminate execution by returning an err
mon.Event("failed_dial") //mon:locked
pingNodeSuccess = false
pingErrorMessage = fmt.Sprintf("failed to dial storage node (ID: %s) at address %s: %q",
nodeurl.ID, nodeurl.Address, err,
)
service.log.Debug("pingBack failed to dial storage node",
zap.String("pingErrorMessage", pingErrorMessage),
)
return pingNodeSuccess, pingNodeSuccessQUIC, pingErrorMessage, nil
}
defer func() { err = errs.Combine(err, client.Close()) }()
_, err = client.pingNode(ctx, &pb.ContactPingRequest{})
if err != nil {
mon.Event("failed_ping_node") //mon:locked
pingNodeSuccess = false
pingErrorMessage = fmt.Sprintf("failed to ping storage node, your node indicated error code: %d, %q", rpcstatus.Code(err), err)
service.log.Debug("pingBack pingNode error",
zap.Stringer("Node ID", nodeurl.ID),
zap.String("pingErrorMessage", pingErrorMessage),
)
return pingNodeSuccess, pingNodeSuccessQUIC, pingErrorMessage, nil
}
pingNodeSuccessQUIC = true
err = service.pingNodeQUIC(ctx, nodeurl)
if err != nil {
// udp ping back is optional right now, it shouldn't affect contact service's
// control flow
pingNodeSuccessQUIC = false
pingErrorMessage = err.Error()
}
return pingNodeSuccess, pingNodeSuccessQUIC, pingErrorMessage, nil
}
func (service *Service) pingNodeQUIC(ctx context.Context, nodeurl storj.NodeURL) error {
udpDialer := service.dialer
udpDialer.Connector = quic.NewDefaultConnector(nil)
udpClient, err := dialNodeURL(ctx, udpDialer, nodeurl)
if err != nil {
mon.Event("failed_dial_quic")
return Error.New("failed to dial storage node (ID: %s) at address %s using QUIC: %q", nodeurl.ID.String(), nodeurl.Address, err)
}
defer func() {
_ = udpClient.Close()
}()
_, err = udpClient.pingNode(ctx, &pb.ContactPingRequest{})
if err != nil {
mon.Event("failed_ping_node_quic")
return Error.New("failed to ping storage node using QUIC, your node indicated error code: %d, %q", rpcstatus.Code(err), err)
}
return nil
}