open-consul/agent/consul/acl_endpoint_legacy.go

259 lines
6.8 KiB
Go
Raw Normal View History

New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
package consul
import (
"fmt"
"time"
"github.com/armon/go-metrics"
"github.com/hashicorp/consul/acl"
"github.com/hashicorp/consul/agent/consul/state"
"github.com/hashicorp/consul/agent/structs"
"github.com/hashicorp/consul/lib"
"github.com/hashicorp/go-memdb"
)
// Bootstrap is used to perform a one-time ACL bootstrap operation on
// a cluster to get the first management token.
func (a *ACL) Bootstrap(args *structs.DCSpecificRequest, reply *structs.ACL) error {
if done, err := a.srv.forward("ACL.Bootstrap", args, args, reply); done {
return err
}
// Verify we are allowed to serve this request
if !a.srv.InACLDatacenter() {
return acl.ErrDisabled
}
// By doing some pre-checks we can head off later bootstrap attempts
// without having to run them through Raft, which should curb abuse.
state := a.srv.fsm.State()
allowed, _, err := state.CanBootstrapACLToken()
if err != nil {
return err
}
if !allowed {
return structs.ACLBootstrapNotAllowedErr
}
// Propose a new token.
token, err := lib.GenerateUUID(a.srv.checkTokenUUID)
if err != nil {
return fmt.Errorf("failed to make random token: %v", err)
}
// Attempt a bootstrap.
req := structs.ACLRequest{
Datacenter: a.srv.config.ACLDatacenter,
Op: structs.ACLBootstrapNow,
ACL: structs.ACL{
ID: token,
Name: "Bootstrap Token",
Type: structs.ACLTokenTypeManagement,
},
}
resp, err := a.srv.raftApply(structs.ACLRequestType, &req)
if err != nil {
return err
}
switch v := resp.(type) {
case error:
return v
case *structs.ACL:
*reply = *v
default:
// Just log this, since it looks like the bootstrap may have
// completed.
a.srv.logger.Printf("[ERR] consul.acl: Unexpected response during bootstrap: %T", v)
}
a.srv.logger.Printf("[INFO] consul.acl: ACL bootstrap completed")
return nil
}
// aclApplyInternal is used to apply an ACL request after it has been vetted that
// this is a valid operation. It is used when users are updating ACLs, in which
// case we check their token to make sure they have management privileges. It is
// also used for ACL replication. We want to run the replicated ACLs through the
// same checks on the change itself.
func aclApplyInternal(srv *Server, args *structs.ACLRequest, reply *string) error {
// All ACLs must have an ID by this point.
if args.ACL.ID == "" {
return fmt.Errorf("Missing ACL ID")
}
switch args.Op {
case structs.ACLSet:
// Verify the ACL type
switch args.ACL.Type {
case structs.ACLTokenTypeClient:
case structs.ACLTokenTypeManagement:
default:
return fmt.Errorf("Invalid ACL Type")
}
_, existing, _ := srv.fsm.State().ACLTokenGetBySecret(nil, args.ACL.ID)
if existing != nil && len(existing.Policies) > 0 {
return fmt.Errorf("Cannot use legacy endpoint to modify a non-legacy token")
}
// Verify this is not a root ACL
if acl.RootAuthorizer(args.ACL.ID) != nil {
return acl.PermissionDeniedError{Cause: "Cannot modify root ACL"}
}
// Validate the rules compile
_, err := acl.NewPolicyFromSource("", 0, args.ACL.Rules, acl.SyntaxLegacy, srv.sentinel)
if err != nil {
return fmt.Errorf("ACL rule compilation failed: %v", err)
}
case structs.ACLDelete:
if args.ACL.ID == anonymousToken {
return acl.PermissionDeniedError{Cause: "Cannot delete anonymous token"}
}
default:
return fmt.Errorf("Invalid ACL Operation")
}
// Apply the update
resp, err := srv.raftApply(structs.ACLRequestType, args)
if err != nil {
srv.logger.Printf("[ERR] consul.acl: Apply failed: %v", err)
return err
}
if respErr, ok := resp.(error); ok {
return respErr
}
// Check if the return type is a string
if respString, ok := resp.(string); ok {
*reply = respString
}
return nil
}
// Apply is used to apply a modifying request to the data store. This should
// only be used for operations that modify the data
func (a *ACL) Apply(args *structs.ACLRequest, reply *string) error {
if done, err := a.srv.forward("ACL.Apply", args, args, reply); done {
return err
}
defer metrics.MeasureSince([]string{"acl", "apply"}, time.Now())
// Verify we are allowed to serve this request
if !a.srv.ACLsEnabled() {
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
return acl.ErrDisabled
}
// Verify token is permitted to modify ACLs
if rule, err := a.srv.ResolveToken(args.Token); err != nil {
return err
} else if rule == nil || !rule.ACLWrite() {
return acl.ErrPermissionDenied
}
// If no ID is provided, generate a new ID. This must be done prior to
// appending to the Raft log, because the ID is not deterministic. Once
// the entry is in the log, the state update MUST be deterministic or
// the followers will not converge.
if args.Op == structs.ACLSet && args.ACL.ID == "" {
var err error
args.ACL.ID, err = lib.GenerateUUID(a.srv.checkTokenUUID)
if err != nil {
return err
}
}
// Do the apply now that this update is vetted.
if err := aclApplyInternal(a.srv, args, reply); err != nil {
return err
}
// Clear the cache if applicable
if args.ACL.ID != "" {
a.srv.acls.cache.RemoveIdentity(args.ACL.ID)
}
return nil
}
// Get is used to retrieve a single ACL
func (a *ACL) Get(args *structs.ACLSpecificRequest,
reply *structs.IndexedACLs) error {
if done, err := a.srv.forward("ACL.Get", args, args, reply); done {
return err
}
// Verify we are allowed to serve this request
if !a.srv.ACLsEnabled() {
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
return acl.ErrDisabled
}
return a.srv.blockingQuery(&args.QueryOptions,
&reply.QueryMeta,
func(ws memdb.WatchSet, state *state.Store) error {
index, token, err := state.ACLTokenGetBySecret(ws, args.ACL)
if err != nil {
return err
}
// converting an ACLToken to an ACL will return nil and an error
// (which we ignore) when it is unconvertible.
var acl *structs.ACL
if token != nil {
acl, _ = token.Convert()
}
reply.Index = index
if acl != nil {
reply.ACLs = structs.ACLs{acl}
} else {
reply.ACLs = nil
}
return nil
})
}
// List is used to list all the ACLs
func (a *ACL) List(args *structs.DCSpecificRequest,
reply *structs.IndexedACLs) error {
if done, err := a.srv.forward("ACL.List", args, args, reply); done {
return err
}
// Verify we are allowed to serve this request
if !a.srv.ACLsEnabled() {
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
return acl.ErrDisabled
}
// Verify token is permitted to list ACLs
if rule, err := a.srv.ResolveToken(args.Token); err != nil {
return err
} else if rule == nil || !rule.ACLWrite() {
return acl.ErrPermissionDenied
}
return a.srv.blockingQuery(&args.QueryOptions,
&reply.QueryMeta,
func(ws memdb.WatchSet, state *state.Store) error {
index, tokens, err := state.ACLTokenList(ws, false, true, "")
if err != nil {
return err
}
var acls structs.ACLs
for _, token := range tokens {
if acl, err := token.Convert(); err == nil && acl != nil {
acls = append(acls, acl)
}
}
reply.Index, reply.ACLs = index, acls
return nil
})
}