open-nomad/nomad/namespace_endpoint.go
Tim Gross 32f6ce1c54
Authenticate method improvements (#15734)
This changeset covers a sidebar discussion that @schmichael and I had around the
design for pre-forwarding auth. This includes some changes extracted out of
#15513 to make it easier to review both and leave a clean history.

* Remove fast path for NodeID. Previously-connected clients will have a NodeID
  set on the context, and because this is a large portion of the RPCs sent we
  fast-pathed it at the top of the `Authenticate` method. But the context is
  shared for all yamux streams over the same yamux session (and TCP
  connection). This lets an authenticated HTTP request to a client use the
  NodeID for authentication, which is a privilege escalation. Remove the fast
  path and annotate it so that we don't break it again.

* Add context to decisions around AuthenticatedIdentity. The `Authenticate`
  method taken on its own looks like it wants to return an `acl.ACL` that folds
  over all the various identity types (creating an ephemeral ACL on the fly if
  neccessary). But keeping these fields idependent allows RPC handlers to
  differentiate between internal and external origins so we most likely want to
  avoid this. Leave some docstrings as a warning as to why this is built the way
  it is.

* Mutate the request rather than returning. When reviewing #15513 we decided
  that forcing the request handler to call `SetIdentity` was repetitive and
  error prone. Instead, the `Authenticate` method mutates the request by setting
  its `AuthenticatedIdentity`.
2023-01-10 09:46:38 -05:00

384 lines
10 KiB
Go

package nomad
import (
"fmt"
"time"
"github.com/armon/go-metrics"
"github.com/hashicorp/go-memdb"
"github.com/hashicorp/go-multierror"
"github.com/hashicorp/nomad/nomad/state"
"github.com/hashicorp/nomad/nomad/structs"
)
// Namespace endpoint is used for manipulating namespaces
type Namespace struct {
srv *Server
ctx *RPCContext
}
func NewNamespaceEndpoint(srv *Server, ctx *RPCContext) *Namespace {
return &Namespace{srv: srv, ctx: ctx}
}
// UpsertNamespaces is used to upsert a set of namespaces
func (n *Namespace) UpsertNamespaces(args *structs.NamespaceUpsertRequest,
reply *structs.GenericResponse) error {
authErr := n.srv.Authenticate(n.ctx, args)
if authErr != nil {
return authErr
}
args.Region = n.srv.config.AuthoritativeRegion
if done, err := n.srv.forward("Namespace.UpsertNamespaces", args, args, reply); done {
return err
}
defer metrics.MeasureSince([]string{"nomad", "namespace", "upsert_namespaces"}, time.Now())
// Check management permissions
if aclObj, err := n.srv.ResolveACL(args.GetIdentity().GetACLToken()); err != nil {
return err
} else if aclObj != nil && !aclObj.IsManagement() {
return structs.ErrPermissionDenied
}
// Validate there is at least one namespace
if len(args.Namespaces) == 0 {
return fmt.Errorf("must specify at least one namespace")
}
// Validate the namespaces and set the hash
for _, ns := range args.Namespaces {
if err := ns.Validate(); err != nil {
return fmt.Errorf("Invalid namespace %q: %v", ns.Name, err)
}
ns.SetHash()
}
// Update via Raft
out, index, err := n.srv.raftApply(structs.NamespaceUpsertRequestType, args)
if err != nil {
return err
}
// Check if there was an error when applying.
if err, ok := out.(error); ok && err != nil {
return err
}
// Update the index
reply.Index = index
return nil
}
// DeleteNamespaces is used to delete a namespace
func (n *Namespace) DeleteNamespaces(args *structs.NamespaceDeleteRequest, reply *structs.GenericResponse) error {
args.Region = n.srv.config.AuthoritativeRegion
if done, err := n.srv.forward("Namespace.DeleteNamespaces", args, args, reply); done {
return err
}
defer metrics.MeasureSince([]string{"nomad", "namespace", "delete_namespaces"}, time.Now())
// Check management permissions
if aclObj, err := n.srv.ResolveToken(args.AuthToken); err != nil {
return err
} else if aclObj != nil && !aclObj.IsManagement() {
return structs.ErrPermissionDenied
}
// Validate at least one namespace
if len(args.Namespaces) == 0 {
return fmt.Errorf("must specify at least one namespace to delete")
}
for _, ns := range args.Namespaces {
if ns == structs.DefaultNamespace {
return fmt.Errorf("can not delete default namespace")
}
}
// Check that the deleting namespaces do not have non-terminal jobs in both
// this region and all federated regions
var mErr multierror.Error
for _, ns := range args.Namespaces {
nonTerminal, err := n.nonTerminalNamespaces(args.AuthToken, ns)
if err != nil {
_ = multierror.Append(&mErr, err)
} else if len(nonTerminal) != 0 {
_ = multierror.Append(&mErr, fmt.Errorf("namespace %q has non-terminal jobs in regions: %v", ns, nonTerminal))
}
}
if err := mErr.ErrorOrNil(); err != nil {
return err
}
// Update via Raft
out, index, err := n.srv.raftApply(structs.NamespaceDeleteRequestType, args)
if err != nil {
return err
}
// Check if there was an error when applying.
if err, ok := out.(error); ok && err != nil {
return err
}
// Update the index
reply.Index = index
return nil
}
// nonTerminalNamespaces returns whether the set of regions in which the
// namespaces contains non-terminal jobs, checking all federated regions
// including this one.
func (n *Namespace) nonTerminalNamespaces(authToken, namespace string) ([]string, error) {
regions := n.srv.Regions()
thisRegion := n.srv.Region()
terminal := make([]string, 0, len(regions))
// Check if this region is terminal
localTerminal, err := n.namespaceTerminalLocally(namespace)
if err != nil {
return nil, err
}
if !localTerminal {
terminal = append(terminal, thisRegion)
}
for _, region := range regions {
if region == thisRegion {
continue
}
remoteTerminal, err := n.namespaceTerminalInRegion(authToken, namespace, region)
if err != nil {
return nil, err
}
if !remoteTerminal {
terminal = append(terminal, region)
}
}
return terminal, nil
}
// namespaceTerminalLocally returns if the namespace contains only terminal jobs
// in the local region .
func (n *Namespace) namespaceTerminalLocally(namespace string) (bool, error) {
snap, err := n.srv.fsm.State().Snapshot()
if err != nil {
return false, err
}
iter, err := snap.JobsByNamespace(nil, namespace)
if err != nil {
return false, err
}
for {
raw := iter.Next()
if raw == nil {
break
}
job := raw.(*structs.Job)
if job.Status != structs.JobStatusDead {
return false, nil
}
}
return true, nil
}
// namespaceTerminalInRegion returns if the namespace contains only terminal
// jobs in the given region .
func (n *Namespace) namespaceTerminalInRegion(authToken, namespace, region string) (bool, error) {
req := &structs.JobListRequest{
QueryOptions: structs.QueryOptions{
Region: region,
Namespace: namespace,
AllowStale: false,
AuthToken: authToken,
},
}
var resp structs.JobListResponse
done, err := n.srv.forward("Job.List", req, req, &resp)
if !done {
return false, fmt.Errorf("unexpectedly did not forward Job.List to region %q", region)
} else if err != nil {
return false, err
}
for _, job := range resp.Jobs {
if job.Status != structs.JobStatusDead {
return false, nil
}
}
return true, nil
}
// ListNamespaces is used to list the namespaces
func (n *Namespace) ListNamespaces(args *structs.NamespaceListRequest, reply *structs.NamespaceListResponse) error {
if done, err := n.srv.forward("Namespace.ListNamespaces", args, args, reply); done {
return err
}
defer metrics.MeasureSince([]string{"nomad", "namespace", "list_namespace"}, time.Now())
// Resolve token to acl to filter namespace list
aclObj, err := n.srv.ResolveToken(args.AuthToken)
if err != nil {
return err
}
// Setup the blocking query
opts := blockingOptions{
queryOpts: &args.QueryOptions,
queryMeta: &reply.QueryMeta,
run: func(ws memdb.WatchSet, s *state.StateStore) error {
// Iterate over all the namespaces
var err error
var iter memdb.ResultIterator
if prefix := args.QueryOptions.Prefix; prefix != "" {
iter, err = s.NamespacesByNamePrefix(ws, prefix)
} else {
iter, err = s.Namespaces(ws)
}
if err != nil {
return err
}
reply.Namespaces = nil
for {
raw := iter.Next()
if raw == nil {
break
}
ns := raw.(*structs.Namespace)
// Only return namespaces allowed by acl
if aclObj == nil || aclObj.AllowNamespace(ns.Name) {
reply.Namespaces = append(reply.Namespaces, ns)
}
}
// Use the last index that affected the namespace table
index, err := s.Index(state.TableNamespaces)
if err != nil {
return err
}
// Ensure we never set the index to zero, otherwise a blocking query cannot be used.
// We floor the index at one, since realistically the first write must have a higher index.
if index == 0 {
index = 1
}
reply.Index = index
return nil
}}
return n.srv.blockingRPC(&opts)
}
// GetNamespace is used to get a specific namespace
func (n *Namespace) GetNamespace(args *structs.NamespaceSpecificRequest, reply *structs.SingleNamespaceResponse) error {
if done, err := n.srv.forward("Namespace.GetNamespace", args, args, reply); done {
return err
}
defer metrics.MeasureSince([]string{"nomad", "namespace", "get_namespace"}, time.Now())
// Check capabilities for the given namespace permissions
if aclObj, err := n.srv.ResolveToken(args.AuthToken); err != nil {
return err
} else if aclObj != nil && !aclObj.AllowNamespace(args.Name) {
return structs.ErrPermissionDenied
}
// Setup the blocking query
opts := blockingOptions{
queryOpts: &args.QueryOptions,
queryMeta: &reply.QueryMeta,
run: func(ws memdb.WatchSet, s *state.StateStore) error {
// Look for the namespace
out, err := s.NamespaceByName(ws, args.Name)
if err != nil {
return err
}
// Setup the output
reply.Namespace = out
if out != nil {
reply.Index = out.ModifyIndex
} else {
// Use the last index that affected the namespace table
index, err := s.Index(state.TableNamespaces)
if err != nil {
return err
}
// Ensure we never set the index to zero, otherwise a blocking query cannot be used.
// We floor the index at one, since realistically the first write must have a higher index.
if index == 0 {
index = 1
}
reply.Index = index
}
return nil
}}
return n.srv.blockingRPC(&opts)
}
// GetNamespaces is used to get a set of namespaces
func (n *Namespace) GetNamespaces(args *structs.NamespaceSetRequest, reply *structs.NamespaceSetResponse) error {
if done, err := n.srv.forward("Namespace.GetNamespaces", args, args, reply); done {
return err
}
defer metrics.MeasureSince([]string{"nomad", "namespace", "get_namespaces"}, time.Now())
// Check management permissions
if aclObj, err := n.srv.ResolveToken(args.AuthToken); err != nil {
return err
} else if aclObj != nil && !aclObj.IsManagement() {
return structs.ErrPermissionDenied
}
// Setup the blocking query
opts := blockingOptions{
queryOpts: &args.QueryOptions,
queryMeta: &reply.QueryMeta,
run: func(ws memdb.WatchSet, s *state.StateStore) error {
// Setup the output
reply.Namespaces = make(map[string]*structs.Namespace, len(args.Namespaces))
// Look for the namespace
for _, namespace := range args.Namespaces {
out, err := s.NamespaceByName(ws, namespace)
if err != nil {
return err
}
if out != nil {
reply.Namespaces[namespace] = out
}
}
// Use the last index that affected the policy table
index, err := s.Index(state.TableNamespaces)
if err != nil {
return err
}
// Ensure we never set the index to zero, otherwise a blocking query cannot be used.
// We floor the index at one, since realistically the first write must have a higher index.
if index == 0 {
index = 1
}
reply.Index = index
return nil
}}
return n.srv.blockingRPC(&opts)
}