mirror of
https://github.com/tailscale/tailscale.git
synced 2026-02-14 09:52:04 -05:00
This file was never truly necessary and has never actually been used in the history of Tailscale's open source releases. A Brief History of AUTHORS files --- The AUTHORS file was a pattern developed at Google, originally for Chromium, then adopted by Go and a bunch of other projects. The problem was that Chromium originally had a copyright line only recognizing Google as the copyright holder. Because Google (and most open source projects) do not require copyright assignemnt for contributions, each contributor maintains their copyright. Some large corporate contributors then tried to add their own name to the copyright line in the LICENSE file or in file headers. This quickly becomes unwieldy, and puts a tremendous burden on anyone building on top of Chromium, since the license requires that they keep all copyright lines intact. The compromise was to create an AUTHORS file that would list all of the copyright holders. The LICENSE file and source file headers would then include that list by reference, listing the copyright holder as "The Chromium Authors". This also become cumbersome to simply keep the file up to date with a high rate of new contributors. Plus it's not always obvious who the copyright holder is. Sometimes it is the individual making the contribution, but many times it may be their employer. There is no way for the proejct maintainer to know. Eventually, Google changed their policy to no longer recommend trying to keep the AUTHORS file up to date proactively, and instead to only add to it when requested: https://opensource.google/docs/releasing/authors. They are also clear that: > Adding contributors to the AUTHORS file is entirely within the > project's discretion and has no implications for copyright ownership. It was primarily added to appease a small number of large contributors that insisted that they be recognized as copyright holders (which was entirely their right to do). But it's not truly necessary, and not even the most accurate way of identifying contributors and/or copyright holders. In practice, we've never added anyone to our AUTHORS file. It only lists Tailscale, so it's not really serving any purpose. It also causes confusion because Tailscalars put the "Tailscale Inc & AUTHORS" header in other open source repos which don't actually have an AUTHORS file, so it's ambiguous what that means. Instead, we just acknowledge that the contributors to Tailscale (whoever they are) are copyright holders for their individual contributions. We also have the benefit of using the DCO (developercertificate.org) which provides some additional certification of their right to make the contribution. The source file changes were purely mechanical with: git ls-files | xargs sed -i -e 's/\(Tailscale Inc &\) AUTHORS/\1 contributors/g' Updates #cleanup Change-Id: Ia101a4a3005adb9118051b3416f5a64a4a45987d Signed-off-by: Will Norris <will@tailscale.com>
190 lines
6.4 KiB
Go
190 lines
6.4 KiB
Go
// Copyright (c) Tailscale Inc & contributors
|
|
// SPDX-License-Identifier: BSD-3-Clause
|
|
|
|
// Package certs implements logic to help multiple Kubernetes replicas share TLS
|
|
// certs for a common Tailscale Service.
|
|
package certs
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"net"
|
|
"slices"
|
|
"sync"
|
|
"time"
|
|
|
|
"tailscale.com/ipn"
|
|
"tailscale.com/kube/localclient"
|
|
"tailscale.com/types/logger"
|
|
"tailscale.com/util/goroutines"
|
|
"tailscale.com/util/mak"
|
|
)
|
|
|
|
// CertManager is responsible for issuing certificates for known domains and for
|
|
// maintaining a loop that re-attempts issuance daily.
|
|
// Currently cert manager logic is only run on ingress ProxyGroup replicas that are responsible for managing certs for
|
|
// HA Ingress HTTPS endpoints ('write' replicas).
|
|
type CertManager struct {
|
|
lc localclient.LocalClient
|
|
logf logger.Logf
|
|
tracker goroutines.Tracker // tracks running goroutines
|
|
mu sync.Mutex // guards the following
|
|
// certLoops contains a map of DNS names, for which we currently need to
|
|
// manage certs to cancel functions that allow stopping a goroutine when
|
|
// we no longer need to manage certs for the DNS name.
|
|
certLoops map[string]context.CancelFunc
|
|
}
|
|
|
|
func NewCertManager(lc localclient.LocalClient, logf logger.Logf) *CertManager {
|
|
return &CertManager{
|
|
lc: lc,
|
|
logf: logf,
|
|
}
|
|
}
|
|
|
|
// EnsureCertLoops ensures that, for all currently managed Service HTTPS
|
|
// endpoints, there is a cert loop responsible for issuing and ensuring the
|
|
// renewal of the TLS certs.
|
|
// ServeConfig must not be nil.
|
|
func (cm *CertManager) EnsureCertLoops(ctx context.Context, sc *ipn.ServeConfig) error {
|
|
if sc == nil {
|
|
return fmt.Errorf("[unexpected] ensureCertLoops called with nil ServeConfig")
|
|
}
|
|
currentDomains := make(map[string]bool)
|
|
const httpsPort = "443"
|
|
for _, service := range sc.Services {
|
|
for hostPort := range service.Web {
|
|
domain, port, err := net.SplitHostPort(string(hostPort))
|
|
if err != nil {
|
|
return fmt.Errorf("[unexpected] unable to parse HostPort %s", hostPort)
|
|
}
|
|
if port != httpsPort { // HA Ingress' HTTP endpoint
|
|
continue
|
|
}
|
|
currentDomains[domain] = true
|
|
}
|
|
}
|
|
cm.mu.Lock()
|
|
defer cm.mu.Unlock()
|
|
for domain := range currentDomains {
|
|
if _, exists := cm.certLoops[domain]; !exists {
|
|
cancelCtx, cancel := context.WithCancel(ctx)
|
|
mak.Set(&cm.certLoops, domain, cancel)
|
|
// Note that most of the issuance anyway happens
|
|
// serially because the cert client has a shared lock
|
|
// that's held during any issuance.
|
|
cm.tracker.Go(func() { cm.runCertLoop(cancelCtx, domain) })
|
|
}
|
|
}
|
|
|
|
// Stop goroutines for domain names that are no longer in the config.
|
|
for domain, cancel := range cm.certLoops {
|
|
if !currentDomains[domain] {
|
|
cancel()
|
|
delete(cm.certLoops, domain)
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
// runCertLoop:
|
|
// - calls localAPI certificate endpoint to ensure that certs are issued for the
|
|
// given domain name
|
|
// - calls localAPI certificate endpoint daily to ensure that certs are renewed
|
|
// - if certificate issuance failed retries after an exponential backoff period
|
|
// starting at 1 minute and capped at 24 hours. Reset the backoff once issuance succeeds.
|
|
// Note that renewal check also happens when the node receives an HTTPS request and it is possible that certs get
|
|
// renewed at that point. Renewal here is needed to prevent the shared certs from expiry in edge cases where the 'write'
|
|
// replica does not get any HTTPS requests.
|
|
// https://letsencrypt.org/docs/integration-guide/#retrying-failures
|
|
func (cm *CertManager) runCertLoop(ctx context.Context, domain string) {
|
|
const (
|
|
normalInterval = 24 * time.Hour // regular renewal check
|
|
initialRetry = 1 * time.Minute // initial backoff after a failure
|
|
maxRetryInterval = 24 * time.Hour // max backoff period
|
|
)
|
|
|
|
if err := cm.waitForCertDomain(ctx, domain); err != nil {
|
|
// Best-effort, log and continue with the issuing loop.
|
|
cm.logf("error waiting for cert domain %s: %v", domain, err)
|
|
}
|
|
|
|
timer := time.NewTimer(0) // fire off timer immediately
|
|
defer timer.Stop()
|
|
retryCount := 0
|
|
for {
|
|
select {
|
|
case <-ctx.Done():
|
|
return
|
|
case <-timer.C:
|
|
// We call the certificate endpoint, but don't do anything with the
|
|
// returned certs here. The call to the certificate endpoint will
|
|
// ensure that certs are issued/renewed as needed and stored in the
|
|
// relevant state store. For example, for HA Ingress 'write' replica,
|
|
// the cert and key will be stored in a Kubernetes Secret named after
|
|
// the domain for which we are issuing.
|
|
//
|
|
// Note that renewals triggered by the call to the certificates
|
|
// endpoint here and by renewal check triggered during a call to
|
|
// node's HTTPS endpoint share the same state/renewal lock mechanism,
|
|
// so we should not run into redundant issuances during concurrent
|
|
// renewal checks.
|
|
|
|
// An issuance holds a shared lock, so we need to avoid a situation
|
|
// where other services cannot issue certs because a single one is
|
|
// holding the lock.
|
|
ctxT, cancel := context.WithTimeout(ctx, time.Second*300)
|
|
_, _, err := cm.lc.CertPair(ctxT, domain)
|
|
cancel()
|
|
if err != nil {
|
|
cm.logf("error refreshing certificate for %s: %v", domain, err)
|
|
}
|
|
var nextInterval time.Duration
|
|
// TODO(irbekrm): distinguish between LE rate limit errors and other
|
|
// error types like transient network errors.
|
|
if err == nil {
|
|
retryCount = 0
|
|
nextInterval = normalInterval
|
|
} else {
|
|
retryCount++
|
|
// Calculate backoff: initialRetry * 2^(retryCount-1)
|
|
// For retryCount=1: 1min * 2^0 = 1min
|
|
// For retryCount=2: 1min * 2^1 = 2min
|
|
// For retryCount=3: 1min * 2^2 = 4min
|
|
backoff := initialRetry * time.Duration(1<<(retryCount-1))
|
|
if backoff > maxRetryInterval {
|
|
backoff = maxRetryInterval
|
|
}
|
|
nextInterval = backoff
|
|
cm.logf("Error refreshing certificate for %s (retry %d): %v. Will retry in %v\n",
|
|
domain, retryCount, err, nextInterval)
|
|
}
|
|
timer.Reset(nextInterval)
|
|
}
|
|
}
|
|
}
|
|
|
|
// waitForCertDomain ensures the requested domain is in the list of allowed
|
|
// domains before issuing the cert for the first time.
|
|
func (cm *CertManager) waitForCertDomain(ctx context.Context, domain string) error {
|
|
w, err := cm.lc.WatchIPNBus(ctx, ipn.NotifyInitialNetMap)
|
|
if err != nil {
|
|
return fmt.Errorf("error watching IPN bus: %w", err)
|
|
}
|
|
defer w.Close()
|
|
|
|
for {
|
|
n, err := w.Next()
|
|
if err != nil {
|
|
return err
|
|
}
|
|
if n.NetMap == nil {
|
|
continue
|
|
}
|
|
|
|
if slices.Contains(n.NetMap.DNS.CertDomains, domain) {
|
|
return nil
|
|
}
|
|
}
|
|
}
|