extern/zrepl
1
0
Fork 0
mirror of https://github.com/zrepl/zrepl.git synced 2025-08-14 09:08:24 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

reimplement io.ReadWriteCloser based RPC mechanism

Browse Source 
The existing ByteStreamRPC requires writing RPC stub + server code
for each RPC endpoint. Does not scale well.

Goal: adding a new RPC call should

- not require writing an RPC stub / handler
- not require modifications to the RPC lib

The wire format is inspired by HTTP2, the API by net/rpc.

Frames are used for framing messages, i.e. a message is made of multiple
frames which are glued together using a frame-bridging reader / writer.
This roughly corresponds to HTTP2 streams, although we're happy with
just one stream at any time and the resulting non-need for flow control,
etc.

Frames are typed using a header. The two most important types are
'Header' and 'Data'.

The RPC protocol is built on top of this:

- Client sends a header         => multiple frames of type 'header'
- Client sends request body     => mulitiple frames of type 'data'
- Server reads a header         => multiple frames of type 'header'
- Server reads request body     => mulitiple frames of type 'data'
- Server sends response header  => ...
- Server sends response body    => ...

An RPC header is serialized JSON and always the same structure.
The body is of the type specified in the header.

The RPC server and client use some semi-fancy reflection tequniques to
automatically infer the data type of the request/response body based on
the method signature of the server handler; or the client parameters,
respectively.
This boils down to a special-case for io.Reader, which are just dumped
into a series of data frames as efficiently as possible.
All other types are (de)serialized using encoding/json.

The RPC layer and Frame Layer log some arbitrary messages that proved
useful during debugging. By default, they log to a non-logger, which
should not have a big impact on performance.

pprof analysis shows the implementation spends its CPU time
        60% waiting for syscalls
        30% in memmove
        10% ...

On a Intel(R) Core(TM) i7-6600U CPU @ 2.60GHz CPU, Linux 4.12, the
implementation achieved ~3.6GiB/s.

Future optimization may include spice(2) / vmspice(2) on Linux, although
this doesn't fit so well with the heavy use of io.Reader / io.Writer
throughout the codebase.

The existing hackaround for local calls was re-implemented to fit the
new interface of PRCServer and RPCClient.
The 'R'PC method invocation is a bit slower because reflection is
involved inbetween, but otherwise performance should be no different.

The RPC code currently does not support multipart requests and thus does
not support the equivalent of a POST.

Thus, the switch to the new rpc code had the following fallout:

- Move request objects + constants from rpc package to main app code
- Sacrifice the hacky 'push = pull me' way of doing push
-> need to further extend RPC to support multipart requests or
     something to implement this properly with additional interfaces
-> should be done after replication is abstracted better than separate
     algorithms for doPull() and doPush()
This commit is contained in:
Christian Schwarz
2017-08-19 22:37:14 +02:00
parent e5b713ce5b
commit 6ab05ee1fa
12 changed files with 938 additions and 815 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
cmd/config.go
View File

@ -32,10 +32,9 @@ type Remote struct {
}
type Transport interface {
Connect(rpcLog Logger) (rpc.RPCRequester, error)
Connect(rpcLog Logger) (rpc.RPCClient, error)
}
type LocalTransport struct {
Handler rpc.RPCHandler
}
type SSHTransport struct {
Host string
@ -53,14 +52,14 @@ type Push struct {
JobName string // for use with jobrun package
To *Remote
Filter zfs.DatasetFilter
InitialReplPolicy rpc.InitialReplPolicy
InitialReplPolicy InitialReplPolicy
RepeatStrategy jobrun.RepeatStrategy
}
type Pull struct {
JobName string // for use with jobrun package
From *Remote
Mapping DatasetMapFilter
InitialReplPolicy rpc.InitialReplPolicy
InitialReplPolicy InitialReplPolicy
RepeatStrategy jobrun.RepeatStrategy
}
@ -161,8 +160,8 @@ func parseRemotes(v interface{}) (remotes map[string]*Remote, err error) {
remotes = make(map[string]*Remote, len(asMap))
for name, p := range asMap {
if name == rpc.LOCAL_TRANSPORT_IDENTITY {
err = errors.New(fmt.Sprintf("remote name '%s' reserved for local pulls", rpc.LOCAL_TRANSPORT_IDENTITY))
if name == LOCAL_TRANSPORT_IDENTITY {
err = errors.New(fmt.Sprintf("remote name '%s' reserved for local pulls", LOCAL_TRANSPORT_IDENTITY))
return
}
@ -238,7 +237,7 @@ func parsePushs(v interface{}, rl remoteLookup) (p map[string]*Push, err error)
return
}
if push.InitialReplPolicy, err = parseInitialReplPolicy(e.InitialReplPolicy, rpc.DEFAULT_INITIAL_REPL_POLICY); err != nil {
if push.InitialReplPolicy, err = parseInitialReplPolicy(e.InitialReplPolicy, DEFAULT_INITIAL_REPL_POLICY); err != nil {
return
}
@ -276,9 +275,9 @@ func parsePulls(v interface{}, rl remoteLookup) (p map[string]*Pull, err error)
var fromRemote *Remote
if e.From == rpc.LOCAL_TRANSPORT_IDENTITY {
if e.From == LOCAL_TRANSPORT_IDENTITY {
fromRemote = &Remote{
Name: rpc.LOCAL_TRANSPORT_IDENTITY,
Name: LOCAL_TRANSPORT_IDENTITY,
Transport: LocalTransport{},
}
} else {
@ -296,7 +295,7 @@ func parsePulls(v interface{}, rl remoteLookup) (p map[string]*Pull, err error)
if pull.Mapping, err = parseDatasetMapFilter(e.Mapping, false); err != nil {
return
}
if pull.InitialReplPolicy, err = parseInitialReplPolicy(e.InitialReplPolicy, rpc.DEFAULT_INITIAL_REPL_POLICY); err != nil {
if pull.InitialReplPolicy, err = parseInitialReplPolicy(e.InitialReplPolicy, DEFAULT_INITIAL_REPL_POLICY); err != nil {
return
}
if pull.RepeatStrategy, err = parseRepeatStrategy(e.Repeat); err != nil {
@ -309,7 +308,7 @@ func parsePulls(v interface{}, rl remoteLookup) (p map[string]*Pull, err error)
return
}
func parseInitialReplPolicy(v interface{}, defaultPolicy rpc.InitialReplPolicy) (p rpc.InitialReplPolicy, err error) {
func parseInitialReplPolicy(v interface{}, defaultPolicy InitialReplPolicy) (p InitialReplPolicy, err error) {
s, ok := v.(string)
if !ok {
goto err
@ -319,9 +318,9 @@ func parseInitialReplPolicy(v interface{}, defaultPolicy rpc.InitialReplPolicy)
case s == "":
p = defaultPolicy
case s == "most_recent":
p = rpc.InitialReplPolicyMostRecent
p = InitialReplPolicyMostRecent
case s == "all":
p = rpc.InitialReplPolicyAll
p = InitialReplPolicyAll
default:
goto err
}
@ -434,7 +433,7 @@ func parseDatasetMapFilter(mi interface{}, filterOnly bool) (f DatasetMapFilter,
return
}
func (t SSHTransport) Connect(rpcLog Logger) (r rpc.RPCRequester, err error) {
func (t SSHTransport) Connect(rpcLog Logger) (r rpc.RPCClient, err error) {
var stream io.ReadWriteCloser
var rpcTransport sshbytestream.SSHTransport
if err = copier.Copy(&rpcTransport, t); err != nil {
@ -447,18 +446,23 @@ func (t SSHTransport) Connect(rpcLog Logger) (r rpc.RPCRequester, err error) {
if err != nil {
return
}
return rpc.ConnectByteStreamRPC(stream, rpcLog)
client := rpc.NewClient(stream)
return client, nil
}
func (t LocalTransport) Connect(rpcLog Logger) (r rpc.RPCRequester, err error) {
if t.Handler == nil {
panic("local transport with uninitialized handler")
func (t LocalTransport) Connect(rpcLog Logger) (r rpc.RPCClient, err error) {
local := rpc.NewLocalRPC()
handler := Handler{
Logger: log,
// Allow access to any dataset since we control what mapping
// is passed to the pull routine.
// All local datasets will be passed to its Map() function,
// but only those for which a mapping exists will actually be pulled.
// We can pay this small performance penalty for now.
PullACL: localPullACL{},
}
return rpc.ConnectLocalRPC(t.Handler), nil
}
func (t *LocalTransport) SetHandler(handler rpc.RPCHandler) {
t.Handler = handler
registerEndpoints(local, handler)
return local, nil
}
func parsePrunes(m interface{}) (rets map[string]*Prune, err error) {