kubo/exchange/bitswap/workers.go
Juan Batiz-Benet 6e705e1ef0 bitswap/provide: improved rate limiting
this PR greatly speeds up providing and add.

(1) Instead of idling workers, we move to a ratelimiter-based worker.
We put this max at 512, so that means _up to_ 512 goroutines. This
is very small load on the node, as each worker is providing to the
dht, which means mostly waiting. It DOES put a large load on the DHT.
but i want to try this out for a while and see if it's a problem.
We can decide later if it is a problem for the network (nothing
stops anyone from re-compiling, but the defaults of course matter).

(2) We add a buffer size for provideKeys, which means that we block
the add process much less. this is a very cheap buffer, as it only
stores keys (it may be even cheaper with a lock + ring buffer
instead of a channel...). This makes add blazing fast-- it was being
rate limited by providing. Add should not be ratelimited by providing
(much, if any) as the user wants to just store the stuff in the local
node's repo. This buffer is initially set to 4096, which means:

  4096 * keysize (~258 bytes + go overhead) ~ 1-1.5MB

this buffer only last a few sec to mins, and is an ok thing to do
for the sake of very fast adds. (this could be a configurable
paramter, certainly for low-mem footprint use cases). At the moment
this is not much, compared to block sizes.

(3) We make the providing EventBegin() + Done(), so that we can
track how long a provide takes, and we can remove workers as they
finish in bsdash and similar tools.

License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
2015-08-04 20:07:25 +02:00

208 lines
5.2 KiB
Go

package bitswap
import (
"time"
process "github.com/ipfs/go-ipfs/Godeps/_workspace/src/github.com/jbenet/goprocess"
ratelimit "github.com/ipfs/go-ipfs/Godeps/_workspace/src/github.com/jbenet/goprocess/ratelimit"
context "github.com/ipfs/go-ipfs/Godeps/_workspace/src/golang.org/x/net/context"
waitable "github.com/ipfs/go-ipfs/thirdparty/waitable"
key "github.com/ipfs/go-ipfs/blocks/key"
eventlog "github.com/ipfs/go-ipfs/thirdparty/eventlog"
)
var TaskWorkerCount = 8
func (bs *Bitswap) startWorkers(px process.Process, ctx context.Context) {
// Start up a worker to handle block requests this node is making
px.Go(func(px process.Process) {
bs.providerConnector(ctx)
})
// Start up workers to handle requests from other nodes for the data on this node
for i := 0; i < TaskWorkerCount; i++ {
i := i
px.Go(func(px process.Process) {
bs.taskWorker(ctx, i)
})
}
// Start up a worker to manage periodically resending our wantlist out to peers
px.Go(func(px process.Process) {
bs.rebroadcastWorker(ctx)
})
// Start up a worker to manage sending out provides messages
px.Go(func(px process.Process) {
bs.provideCollector(ctx)
})
// Spawn up multiple workers to handle incoming blocks
// consider increasing number if providing blocks bottlenecks
// file transfers
px.Go(bs.provideWorker)
}
func (bs *Bitswap) taskWorker(ctx context.Context, id int) {
idmap := eventlog.LoggableMap{"ID": id}
defer log.Info("bitswap task worker shutting down...")
for {
log.Event(ctx, "Bitswap.TaskWorker.Loop", idmap)
select {
case nextEnvelope := <-bs.engine.Outbox():
select {
case envelope, ok := <-nextEnvelope:
if !ok {
continue
}
log.Event(ctx, "Bitswap.TaskWorker.Work", eventlog.LoggableMap{
"ID": id,
"Target": envelope.Peer.Pretty(),
"Block": envelope.Block.Multihash.B58String(),
})
bs.wm.SendBlock(ctx, envelope)
case <-ctx.Done():
return
}
case <-ctx.Done():
return
}
}
}
func (bs *Bitswap) provideWorker(px process.Process) {
limiter := ratelimit.NewRateLimiter(px, provideWorkerMax)
limitedGoProvide := func(k key.Key, wid int) {
ev := eventlog.LoggableMap{"ID": wid}
limiter.LimitedGo(func(px process.Process) {
ctx := waitable.Context(px) // derive ctx from px
defer log.EventBegin(ctx, "Bitswap.ProvideWorker.Work", ev, &k).Done()
ctx, cancel := context.WithTimeout(ctx, provideTimeout) // timeout ctx
defer cancel()
if err := bs.network.Provide(ctx, k); err != nil {
log.Error(err)
}
})
}
// worker spawner, reads from bs.provideKeys until it closes, spawning a
// _ratelimited_ number of workers to handle each key.
limiter.Go(func(px process.Process) {
for wid := 2; ; wid++ {
ev := eventlog.LoggableMap{"ID": 1}
log.Event(waitable.Context(px), "Bitswap.ProvideWorker.Loop", ev)
select {
case <-px.Closing():
return
case k, ok := <-bs.provideKeys:
if !ok {
log.Debug("provideKeys channel closed")
return
}
limitedGoProvide(k, wid)
}
}
})
}
func (bs *Bitswap) provideCollector(ctx context.Context) {
defer close(bs.provideKeys)
var toProvide []key.Key
var nextKey key.Key
var keysOut chan key.Key
for {
select {
case blk, ok := <-bs.newBlocks:
if !ok {
log.Debug("newBlocks channel closed")
return
}
if keysOut == nil {
nextKey = blk.Key()
keysOut = bs.provideKeys
} else {
toProvide = append(toProvide, blk.Key())
}
case keysOut <- nextKey:
if len(toProvide) > 0 {
nextKey = toProvide[0]
toProvide = toProvide[1:]
} else {
keysOut = nil
}
case <-ctx.Done():
return
}
}
}
// connects to providers for the given keys
func (bs *Bitswap) providerConnector(parent context.Context) {
defer log.Info("bitswap client worker shutting down...")
for {
log.Event(parent, "Bitswap.ProviderConnector.Loop")
select {
case req := <-bs.findKeys:
keys := req.keys
if len(keys) == 0 {
log.Warning("Received batch request for zero blocks")
continue
}
log.Event(parent, "Bitswap.ProviderConnector.Work", eventlog.LoggableMap{"Keys": keys})
// NB: Optimization. Assumes that providers of key[0] are likely to
// be able to provide for all keys. This currently holds true in most
// every situation. Later, this assumption may not hold as true.
child, cancel := context.WithTimeout(req.ctx, providerRequestTimeout)
providers := bs.network.FindProvidersAsync(child, keys[0], maxProvidersPerRequest)
for p := range providers {
go bs.network.ConnectTo(req.ctx, p)
}
cancel()
case <-parent.Done():
return
}
}
}
func (bs *Bitswap) rebroadcastWorker(parent context.Context) {
ctx, cancel := context.WithCancel(parent)
defer cancel()
broadcastSignal := time.NewTicker(rebroadcastDelay.Get())
defer broadcastSignal.Stop()
tick := time.NewTicker(10 * time.Second)
defer tick.Stop()
for {
log.Event(ctx, "Bitswap.Rebroadcast.idle")
select {
case <-tick.C:
n := bs.wm.wl.Len()
if n > 0 {
log.Debug(n, "keys in bitswap wantlist")
}
case <-broadcastSignal.C: // resend unfulfilled wantlist keys
log.Event(ctx, "Bitswap.Rebroadcast.active")
entries := bs.wm.wl.Entries()
if len(entries) > 0 {
bs.connectToProviders(ctx, entries)
}
case <-parent.Done():
return
}
}
}