go-ethereum/p2p/dnsdisc/sync.go
Felix Lange d36276d85e
p2p/dnsdisc: fix hot-spin when all trees are empty (#22313)
In the random sync algorithm used by the DNS node iterator, we first pick a random
tree and then perform one sync action on that tree. This happens in a loop until any
node is found. If no trees contain any nodes, the iterator will enter a hot loop spinning
at 100% CPU.

The fix is complicated. The iterator now checks if a meaningful sync action can
be performed on any tree. If there is nothing to do, it waits for the next root record
recheck time to arrive and then tries again.

Fixes #22306
2021-02-19 09:54:46 +01:00

330 lines
8.5 KiB
Go

// Copyright 2019 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
package dnsdisc
import (
"context"
"math/rand"
"time"
"github.com/ethereum/go-ethereum/common/mclock"
"github.com/ethereum/go-ethereum/p2p/enode"
)
// This is the number of consecutive leaf requests that may fail before
// we consider re-resolving the tree root.
const rootRecheckFailCount = 5
// clientTree is a full tree being synced.
type clientTree struct {
c *Client
loc *linkEntry // link to this tree
lastRootCheck mclock.AbsTime // last revalidation of root
leafFailCount int
rootFailCount int
root *rootEntry
enrs *subtreeSync
links *subtreeSync
lc *linkCache // tracks all links between all trees
curLinks map[string]struct{} // links contained in this tree
linkGCRoot string // root on which last link GC has run
}
func newClientTree(c *Client, lc *linkCache, loc *linkEntry) *clientTree {
return &clientTree{c: c, lc: lc, loc: loc}
}
// syncAll retrieves all entries of the tree.
func (ct *clientTree) syncAll(dest map[string]entry) error {
if err := ct.updateRoot(context.Background()); err != nil {
return err
}
if err := ct.links.resolveAll(dest); err != nil {
return err
}
if err := ct.enrs.resolveAll(dest); err != nil {
return err
}
return nil
}
// syncRandom retrieves a single entry of the tree. The Node return value
// is non-nil if the entry was a node.
func (ct *clientTree) syncRandom(ctx context.Context) (n *enode.Node, err error) {
if ct.rootUpdateDue() {
if err := ct.updateRoot(ctx); err != nil {
return nil, err
}
}
// Update fail counter for leaf request errors.
defer func() {
if err != nil {
ct.leafFailCount++
}
}()
// Link tree sync has priority, run it to completion before syncing ENRs.
if !ct.links.done() {
err := ct.syncNextLink(ctx)
return nil, err
}
ct.gcLinks()
// Sync next random entry in ENR tree. Once every node has been visited, we simply
// start over. This is fine because entries are cached internally by the client LRU
// also by DNS resolvers.
if ct.enrs.done() {
ct.enrs = newSubtreeSync(ct.c, ct.loc, ct.root.eroot, false)
}
return ct.syncNextRandomENR(ctx)
}
// canSyncRandom checks if any meaningful action can be performed by syncRandom.
func (ct *clientTree) canSyncRandom() bool {
// Note: the check for non-zero leaf count is very important here.
// If we're done syncing all nodes, and no leaves were found, the tree
// is empty and we can't use it for sync.
return ct.rootUpdateDue() || !ct.links.done() || !ct.enrs.done() || ct.enrs.leaves != 0
}
// gcLinks removes outdated links from the global link cache. GC runs once
// when the link sync finishes.
func (ct *clientTree) gcLinks() {
if !ct.links.done() || ct.root.lroot == ct.linkGCRoot {
return
}
ct.lc.resetLinks(ct.loc.str, ct.curLinks)
ct.linkGCRoot = ct.root.lroot
}
func (ct *clientTree) syncNextLink(ctx context.Context) error {
hash := ct.links.missing[0]
e, err := ct.links.resolveNext(ctx, hash)
if err != nil {
return err
}
ct.links.missing = ct.links.missing[1:]
if dest, ok := e.(*linkEntry); ok {
ct.lc.addLink(ct.loc.str, dest.str)
ct.curLinks[dest.str] = struct{}{}
}
return nil
}
func (ct *clientTree) syncNextRandomENR(ctx context.Context) (*enode.Node, error) {
index := rand.Intn(len(ct.enrs.missing))
hash := ct.enrs.missing[index]
e, err := ct.enrs.resolveNext(ctx, hash)
if err != nil {
return nil, err
}
ct.enrs.missing = removeHash(ct.enrs.missing, index)
if ee, ok := e.(*enrEntry); ok {
return ee.node, nil
}
return nil, nil
}
func (ct *clientTree) String() string {
return ct.loc.String()
}
// removeHash removes the element at index from h.
func removeHash(h []string, index int) []string {
if len(h) == 1 {
return nil
}
last := len(h) - 1
if index < last {
h[index] = h[last]
h[last] = ""
}
return h[:last]
}
// updateRoot ensures that the given tree has an up-to-date root.
func (ct *clientTree) updateRoot(ctx context.Context) error {
if !ct.slowdownRootUpdate(ctx) {
return ctx.Err()
}
ct.lastRootCheck = ct.c.clock.Now()
ctx, cancel := context.WithTimeout(ctx, ct.c.cfg.Timeout)
defer cancel()
root, err := ct.c.resolveRoot(ctx, ct.loc)
if err != nil {
ct.rootFailCount++
return err
}
ct.root = &root
ct.rootFailCount = 0
ct.leafFailCount = 0
// Invalidate subtrees if changed.
if ct.links == nil || root.lroot != ct.links.root {
ct.links = newSubtreeSync(ct.c, ct.loc, root.lroot, true)
ct.curLinks = make(map[string]struct{})
}
if ct.enrs == nil || root.eroot != ct.enrs.root {
ct.enrs = newSubtreeSync(ct.c, ct.loc, root.eroot, false)
}
return nil
}
// rootUpdateDue returns true when a root update is needed.
func (ct *clientTree) rootUpdateDue() bool {
tooManyFailures := ct.leafFailCount > rootRecheckFailCount
scheduledCheck := ct.c.clock.Now() >= ct.nextScheduledRootCheck()
return ct.root == nil || tooManyFailures || scheduledCheck
}
func (ct *clientTree) nextScheduledRootCheck() mclock.AbsTime {
return ct.lastRootCheck.Add(ct.c.cfg.RecheckInterval)
}
// slowdownRootUpdate applies a delay to root resolution if is tried
// too frequently. This avoids busy polling when the client is offline.
// Returns true if the timeout passed, false if sync was canceled.
func (ct *clientTree) slowdownRootUpdate(ctx context.Context) bool {
var delay time.Duration
switch {
case ct.rootFailCount > 20:
delay = 10 * time.Second
case ct.rootFailCount > 5:
delay = 5 * time.Second
default:
return true
}
timeout := ct.c.clock.NewTimer(delay)
defer timeout.Stop()
select {
case <-timeout.C():
return true
case <-ctx.Done():
return false
}
}
// subtreeSync is the sync of an ENR or link subtree.
type subtreeSync struct {
c *Client
loc *linkEntry
root string
missing []string // missing tree node hashes
link bool // true if this sync is for the link tree
leaves int // counter of synced leaves
}
func newSubtreeSync(c *Client, loc *linkEntry, root string, link bool) *subtreeSync {
return &subtreeSync{c, loc, root, []string{root}, link, 0}
}
func (ts *subtreeSync) done() bool {
return len(ts.missing) == 0
}
func (ts *subtreeSync) resolveAll(dest map[string]entry) error {
for !ts.done() {
hash := ts.missing[0]
ctx, cancel := context.WithTimeout(context.Background(), ts.c.cfg.Timeout)
e, err := ts.resolveNext(ctx, hash)
cancel()
if err != nil {
return err
}
dest[hash] = e
ts.missing = ts.missing[1:]
}
return nil
}
func (ts *subtreeSync) resolveNext(ctx context.Context, hash string) (entry, error) {
e, err := ts.c.resolveEntry(ctx, ts.loc.domain, hash)
if err != nil {
return nil, err
}
switch e := e.(type) {
case *enrEntry:
if ts.link {
return nil, errENRInLinkTree
}
ts.leaves++
case *linkEntry:
if !ts.link {
return nil, errLinkInENRTree
}
ts.leaves++
case *branchEntry:
ts.missing = append(ts.missing, e.children...)
}
return e, nil
}
// linkCache tracks links between trees.
type linkCache struct {
backrefs map[string]map[string]struct{}
changed bool
}
func (lc *linkCache) isReferenced(r string) bool {
return len(lc.backrefs[r]) != 0
}
func (lc *linkCache) addLink(from, to string) {
if _, ok := lc.backrefs[to][from]; ok {
return
}
if lc.backrefs == nil {
lc.backrefs = make(map[string]map[string]struct{})
}
if _, ok := lc.backrefs[to]; !ok {
lc.backrefs[to] = make(map[string]struct{})
}
lc.backrefs[to][from] = struct{}{}
lc.changed = true
}
// resetLinks clears all links of the given tree.
func (lc *linkCache) resetLinks(from string, keep map[string]struct{}) {
stk := []string{from}
for len(stk) > 0 {
item := stk[len(stk)-1]
stk = stk[:len(stk)-1]
for r, refs := range lc.backrefs {
if _, ok := keep[r]; ok {
continue
}
if _, ok := refs[item]; !ok {
continue
}
lc.changed = true
delete(refs, item)
if len(refs) == 0 {
delete(lc.backrefs, r)
stk = append(stk, r)
}
}
}
}