open-consul/agent/consul/operator_raft_endpoint.go

186 lines
5.5 KiB
Go
Raw Normal View History

package consul
import (
"fmt"
"net"
"github.com/hashicorp/raft"
"github.com/hashicorp/serf/serf"
"github.com/hashicorp/consul/acl"
"github.com/hashicorp/consul/agent/metadata"
"github.com/hashicorp/consul/agent/structs"
)
// RaftGetConfiguration is used to retrieve the current Raft configuration.
func (op *Operator) RaftGetConfiguration(args *structs.DCSpecificRequest, reply *structs.RaftConfigurationResponse) error {
if done, err := op.srv.ForwardRPC("Operator.RaftGetConfiguration", args, reply); done {
return err
}
// This action requires operator read access.
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
rule, err := op.srv.ResolveToken(args.Token)
if err != nil {
return err
}
if rule != nil && rule.OperatorRead(nil) != acl.Allow {
return acl.ErrPermissionDenied
}
// We can't fetch the leader and the configuration atomically with
// the current Raft API.
future := op.srv.raft.GetConfiguration()
if err := future.Error(); err != nil {
return err
}
// Index the Consul information about the servers.
serverMap := make(map[raft.ServerAddress]serf.Member)
for _, member := range op.srv.serfLAN.Members() {
valid, parts := metadata.IsConsulServer(member)
if !valid {
continue
}
addr := (&net.TCPAddr{IP: member.Addr, Port: parts.Port}).String()
serverMap[raft.ServerAddress(addr)] = member
}
// Fill out the reply.
leader := op.srv.raft.Leader()
reply.Index = future.Index()
for _, server := range future.Configuration().Servers {
node := "(unknown)"
raftProtocolVersion := "unknown"
if member, ok := serverMap[server.Address]; ok {
node = member.Name
raftProtocolVersion = member.Tags["raft_vsn"]
}
entry := &structs.RaftServer{
ID: server.ID,
Node: node,
Address: server.Address,
Leader: server.Address == leader,
Voter: server.Suffrage == raft.Voter,
ProtocolVersion: raftProtocolVersion,
}
reply.Servers = append(reply.Servers, entry)
}
return nil
}
// RaftRemovePeerByAddress is used to kick a stale peer (one that it in the Raft
// quorum but no longer known to Serf or the catalog) by address in the form of
// "IP:port". The reply argument is not used, but it required to fulfill the RPC
// interface.
func (op *Operator) RaftRemovePeerByAddress(args *structs.RaftRemovePeerRequest, reply *struct{}) error {
if done, err := op.srv.ForwardRPC("Operator.RaftRemovePeerByAddress", args, reply); done {
return err
}
// This is a super dangerous operation that requires operator write
// access.
identity, rule, err := op.srv.ResolveTokenToIdentityAndAuthorizer(args.Token)
if err != nil {
return err
}
if err := op.srv.validateEnterpriseToken(identity); err != nil {
return err
}
if rule != nil && rule.OperatorWrite(nil) != acl.Allow {
return acl.ErrPermissionDenied
}
// Since this is an operation designed for humans to use, we will return
// an error if the supplied address isn't among the peers since it's
// likely they screwed up.
{
future := op.srv.raft.GetConfiguration()
if err := future.Error(); err != nil {
return err
}
for _, s := range future.Configuration().Servers {
if s.Address == args.Address {
args.ID = s.ID
goto REMOVE
}
}
return fmt.Errorf("address %q was not found in the Raft configuration",
args.Address)
}
REMOVE:
if err := op.srv.autopilot.RemoveServer(args.ID); err != nil {
op.logger.Warn("Failed to remove Raft server",
"address", args.Address,
"error", err,
)
return err
}
op.logger.Warn("Removed Raft server", "address", args.Address)
return nil
}
// RaftRemovePeerByID is used to kick a stale peer (one that is in the Raft
// quorum but no longer known to Serf or the catalog) by address in the form of
// "IP:port". The reply argument is not used, but is required to fulfill the RPC
// interface.
func (op *Operator) RaftRemovePeerByID(args *structs.RaftRemovePeerRequest, reply *struct{}) error {
if done, err := op.srv.ForwardRPC("Operator.RaftRemovePeerByID", args, reply); done {
return err
}
// This is a super dangerous operation that requires operator write
// access.
identity, rule, err := op.srv.ResolveTokenToIdentityAndAuthorizer(args.Token)
if err != nil {
return err
}
if err := op.srv.validateEnterpriseToken(identity); err != nil {
return err
}
if rule != nil && rule.OperatorWrite(nil) != acl.Allow {
return acl.ErrPermissionDenied
}
// Since this is an operation designed for humans to use, we will return
// an error if the supplied id isn't among the peers since it's
// likely they screwed up.
{
future := op.srv.raft.GetConfiguration()
if err := future.Error(); err != nil {
return err
}
for _, s := range future.Configuration().Servers {
if s.ID == args.ID {
args.Address = s.Address
goto REMOVE
}
}
return fmt.Errorf("id %q was not found in the Raft configuration",
args.ID)
}
REMOVE:
// The Raft library itself will prevent various forms of foot-shooting,
// like making a configuration with no voters. Some consideration was
// given here to adding more checks, but it was decided to make this as
// low-level and direct as possible. We've got ACL coverage to lock this
// down, and if you are an operator, it's assumed you know what you are
// doing if you are calling this. If you remove a peer that's known to
// Serf, for example, it will come back when the leader does a reconcile
// pass.
if err := op.srv.autopilot.RemoveServer(args.ID); err != nil {
op.logger.Warn("Failed to remove Raft peer with id",
"peer_id", args.ID,
"error", err,
)
return err
}
op.logger.Warn("Removed Raft peer with id", "peer_id", args.ID)
return nil
}