Compare commits
13 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
| 593604dfdc | |||
| b8f888f864 | |||
| 192b2ae621 | |||
| b7f8cb5094 | |||
| a23da6eb57 | |||
| 4c3aa40564 | |||
| 84e2c07a7e | |||
| 680af28bcc | |||
| d94db42ffe | |||
| 93cd83c55c | |||
| 5565fca3ac | |||
| d625ab8d92 | |||
| a3f82c140b |
+13
-1
@@ -15,4 +15,16 @@ builds:
|
||||
- goos: freebsd
|
||||
goarch: arm64
|
||||
- goos: windows
|
||||
goarch: arm64
|
||||
goarch: arm64
|
||||
|
||||
# use zip format for windows
|
||||
archives:
|
||||
- id: default
|
||||
format: tar.gz
|
||||
name_template: "{{ .ProjectName }}_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
|
||||
builds_info:
|
||||
group: root
|
||||
owner: root
|
||||
format_overrides:
|
||||
- goos: windows
|
||||
format: zip
|
||||
@@ -1,4 +1,7 @@
|
||||

|
||||

|
||||

|
||||

|
||||
|
||||
# llama-swap
|
||||
|
||||
@@ -64,13 +67,20 @@ models:
|
||||
# Default (and minimum) is 15 seconds
|
||||
healthCheckTimeout: 60
|
||||
|
||||
# Write HTTP logs (useful for troubleshooting), defaults to false
|
||||
logRequests: true
|
||||
# Valid log levels: debug, info (default), warn, error
|
||||
logLevel: info
|
||||
|
||||
# define valid model values and the upstream server start
|
||||
models:
|
||||
"llama":
|
||||
cmd: llama-server --port 8999 -m Llama-3.2-1B-Instruct-Q4_K_M.gguf
|
||||
# multiline for readability
|
||||
cmd: >
|
||||
llama-server --port 8999
|
||||
--model path/to/Qwen2.5-1.5B-Instruct-Q4_K_M.gguf
|
||||
|
||||
# environment variables to pass to the command
|
||||
env:
|
||||
- "CUDA_VISIBLE_DEVICES=0"
|
||||
|
||||
# where to reach the server started by cmd, make sure the ports match
|
||||
proxy: http://127.0.0.1:8999
|
||||
@@ -91,16 +101,9 @@ models:
|
||||
# default: 0 = never unload model
|
||||
ttl: 60
|
||||
|
||||
"qwen":
|
||||
# environment variables to pass to the command
|
||||
env:
|
||||
- "CUDA_VISIBLE_DEVICES=0"
|
||||
|
||||
# multiline for readability
|
||||
cmd: >
|
||||
llama-server --port 8999
|
||||
--model path/to/Qwen2.5-1.5B-Instruct-Q4_K_M.gguf
|
||||
proxy: http://127.0.0.1:8999
|
||||
# `useModelName` overrides the model name in the request
|
||||
# and sends a specific name to the upstream server
|
||||
useModelName: "qwen:qwq"
|
||||
|
||||
# unlisted models do not show up in /v1/models or /upstream lists
|
||||
# but they can still be requested as normal
|
||||
@@ -117,14 +120,7 @@ models:
|
||||
ghcr.io/ggerganov/llama.cpp:server
|
||||
--model '/models/Qwen2.5-Coder-0.5B-Instruct-Q4_K_M.gguf'
|
||||
|
||||
# `useModelName` will send a specific model name to the upstream server
|
||||
# overriding whatever was set in the request
|
||||
"qwq":
|
||||
proxy: http://127.0.0.1:11434
|
||||
cmd: my-server
|
||||
useModelName: "qwen:qwq"
|
||||
|
||||
# profiles make it easy to managing multi model (and gpu) configurations.
|
||||
# profiles eliminates swapping by running multiple models at the same time
|
||||
#
|
||||
# Tips:
|
||||
# - each model must be listening on a unique address and port
|
||||
@@ -132,8 +128,8 @@ models:
|
||||
# - the profile will load and unload all models in the profile at the same time
|
||||
profiles:
|
||||
coding:
|
||||
- "qwen"
|
||||
- "llama"
|
||||
- "qwen-unlisted"
|
||||
```
|
||||
|
||||
### Use Case Examples
|
||||
@@ -223,9 +219,15 @@ Of course, CLI access is also supported:
|
||||
# sends up to the last 10KB of logs
|
||||
curl http://host/logs'
|
||||
|
||||
# streams logs
|
||||
# streams combined logs
|
||||
curl -Ns 'http://host/logs/stream'
|
||||
|
||||
# just llama-swap's logs
|
||||
curl -Ns 'http://host/logs/stream/proxy'
|
||||
|
||||
# just upstream's logs
|
||||
curl -Ns 'http://host/logs/stream/upstream'
|
||||
|
||||
# stream and filter logs with linux pipes
|
||||
curl -Ns http://host/logs/stream | grep 'eval time'
|
||||
|
||||
|
||||
+3
-3
@@ -1,9 +1,9 @@
|
||||
# Seconds to wait for llama.cpp to be available to serve requests
|
||||
# Default (and minimum): 15 seconds
|
||||
healthCheckTimeout: 15
|
||||
healthCheckTimeout: 90
|
||||
|
||||
# Log HTTP requests helpful for troubleshoot, defaults to False
|
||||
logRequests: true
|
||||
# valid log levels: debug, info (default), warn, error
|
||||
logLevel: info
|
||||
|
||||
models:
|
||||
"llama":
|
||||
|
||||
@@ -27,6 +27,7 @@ func (m *ModelConfig) SanitizedCommand() ([]string, error) {
|
||||
type Config struct {
|
||||
HealthCheckTimeout int `yaml:"healthCheckTimeout"`
|
||||
LogRequests bool `yaml:"logRequests"`
|
||||
LogLevel string `yaml:"logLevel"`
|
||||
Models map[string]ModelConfig `yaml:"models"`
|
||||
Profiles map[string][]string `yaml:"profiles"`
|
||||
|
||||
|
||||
+188
-75
@@ -12,32 +12,65 @@
|
||||
flex-direction: column;
|
||||
font-family: "Courier New", Courier, monospace;
|
||||
}
|
||||
#log-controls {
|
||||
margin: 0.5em;
|
||||
.log-container {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
justify-content: space-between; /* Spaces out elements evenly */
|
||||
}
|
||||
#log-controls input {
|
||||
flex: 1;
|
||||
}
|
||||
#log-controls input:focus {
|
||||
outline: none; /* Ensures no outline is shown when the input is focused */
|
||||
}
|
||||
#log-stream {
|
||||
flex: 1;
|
||||
gap: 0.5em;
|
||||
margin: 0.5em;
|
||||
min-height: 0;
|
||||
}
|
||||
.log-column {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
flex: 1;
|
||||
min-width: 0;
|
||||
transition: flex 0.3s ease;
|
||||
}
|
||||
.log-column.minimized {
|
||||
flex: 0.1;
|
||||
max-width: 50px;
|
||||
border: 1px solid #777;
|
||||
color: green;
|
||||
}
|
||||
.log-controls {
|
||||
display: grid;
|
||||
grid-template-columns: 1fr auto;
|
||||
gap: 0.5em;
|
||||
margin-bottom: 0.5em;
|
||||
}
|
||||
.log-controls input {
|
||||
width: 100%;
|
||||
padding: 4px;
|
||||
}
|
||||
.log-controls input:focus {
|
||||
outline: none;
|
||||
}
|
||||
.log-stream {
|
||||
flex: 1;
|
||||
padding: 1em;
|
||||
background: #f4f4f4;
|
||||
overflow-y: auto;
|
||||
white-space: pre-wrap; /* Ensures line wrapping */
|
||||
word-wrap: break-word; /* Ensures long words wrap */
|
||||
white-space: pre-wrap;
|
||||
word-wrap: break-word;
|
||||
min-height: 0;
|
||||
}
|
||||
|
||||
.regex-error {
|
||||
background-color: #ff0000 !important;
|
||||
}
|
||||
|
||||
/* Make headers clickable and show pointer cursor */
|
||||
h2 {
|
||||
cursor: pointer;
|
||||
user-select: none;
|
||||
margin: 0 0 0.5em 0;
|
||||
padding: 0.5em;
|
||||
}
|
||||
|
||||
h2:hover {
|
||||
background-color: rgba(0, 0, 0, 0.05);
|
||||
}
|
||||
|
||||
/* Dark mode styles */
|
||||
@media (prefers-color-scheme: dark) {
|
||||
body {
|
||||
@@ -45,101 +78,181 @@
|
||||
color: #fff;
|
||||
}
|
||||
|
||||
#log-stream {
|
||||
.log-stream {
|
||||
background: #444;
|
||||
color: #fff;
|
||||
}
|
||||
|
||||
#log-controls input {
|
||||
.log-controls input {
|
||||
background: #555;
|
||||
color: #fff;
|
||||
border: 1px solid #777;
|
||||
}
|
||||
|
||||
#log-controls button {
|
||||
.log-controls button {
|
||||
background: #555;
|
||||
color: #fff;
|
||||
border: 1px solid #777;
|
||||
}
|
||||
|
||||
h2:hover {
|
||||
background-color: rgba(255, 255, 255, 0.1);
|
||||
}
|
||||
}
|
||||
|
||||
/* Hide content when minimized */
|
||||
.log-column.minimized .log-controls,
|
||||
.log-column.minimized .log-stream {
|
||||
display: none;
|
||||
}
|
||||
|
||||
.log-column.minimized h2 {
|
||||
writing-mode: vertical-rl;
|
||||
text-orientation: mixed;
|
||||
transform: rotate(180deg);
|
||||
white-space: nowrap;
|
||||
margin: auto;
|
||||
}
|
||||
</style>
|
||||
</head>
|
||||
<body>
|
||||
<pre id="log-stream">Waiting for logs...</pre>
|
||||
<div id="log-controls">
|
||||
<input type="text" id="filter-input" placeholder="regex filter">
|
||||
<button id="clear-button">clear</button>
|
||||
<div class="log-container">
|
||||
<div class="log-column">
|
||||
<h2>Proxy Logs</h2>
|
||||
<div class="log-controls">
|
||||
<input type="text" id="proxy-filter-input" placeholder="proxy regex filter">
|
||||
<button id="proxy-clear-button">clear</button>
|
||||
</div>
|
||||
<pre class="log-stream" id="proxy-log-stream">Waiting for proxy logs...</pre>
|
||||
</div>
|
||||
<div class="log-column minimized">
|
||||
<h2>Upstream Logs</h2>
|
||||
<div class="log-controls">
|
||||
<input type="text" id="upstream-filter-input" placeholder="upstream regex filter">
|
||||
<button id="upstream-clear-button">clear</button>
|
||||
</div>
|
||||
<pre class="log-stream" id="upstream-log-stream">Waiting for upstream logs...</pre>
|
||||
</div>
|
||||
</div>
|
||||
<script>
|
||||
const logStream = document.getElementById('log-stream');
|
||||
const filterInput = document.getElementById('filter-input');
|
||||
var logData = "";
|
||||
let regexFilter = null;
|
||||
class LogStream {
|
||||
constructor(streamElement, filterInput, clearButton, endpoint) {
|
||||
this.streamElement = streamElement;
|
||||
this.filterInput = filterInput;
|
||||
this.clearButton = clearButton;
|
||||
this.endpoint = endpoint;
|
||||
this.logData = "";
|
||||
this.regexFilter = null;
|
||||
this.eventSource = null;
|
||||
|
||||
function setupEventSource() {
|
||||
if (typeof(EventSource) !== "undefined") {
|
||||
const eventSource = new EventSource("/logs/streamSSE");
|
||||
|
||||
eventSource.onmessage = function(event) {
|
||||
logData += event.data;
|
||||
render()
|
||||
};
|
||||
|
||||
eventSource.onerror = function(err) {
|
||||
logData = "EventSource failed: " + err.message;
|
||||
};
|
||||
} else {
|
||||
logData = "SSE Not supported by this browser."
|
||||
this.initialize();
|
||||
}
|
||||
}
|
||||
|
||||
// poor-ai's react ¯\_(ツ)_/¯
|
||||
function render() {
|
||||
if (regexFilter) {
|
||||
const lines = logData.split('\n');
|
||||
const filteredLines = lines.filter(line => {
|
||||
return regexFilter === null || regexFilter.test(line);
|
||||
initialize() {
|
||||
this.filterInput.addEventListener('input', () => this.updateFilter());
|
||||
this.clearButton.addEventListener('click', () => {
|
||||
this.filterInput.value = "";
|
||||
this.regexFilter = null;
|
||||
this.render();
|
||||
});
|
||||
|
||||
if (filteredLines.length > 0) {
|
||||
logStream.textContent = filteredLines.join('\n') + '\n';
|
||||
} else {
|
||||
logStream.textContent = "";
|
||||
}
|
||||
} else {
|
||||
logStream.textContent = logData;
|
||||
this.setupEventSource();
|
||||
}
|
||||
|
||||
logStream.scrollTop = logStream.scrollHeight;
|
||||
}
|
||||
setupEventSource() {
|
||||
if (typeof(EventSource) === "undefined") {
|
||||
this.logData = "SSE Not supported by this browser.";
|
||||
this.render();
|
||||
return;
|
||||
}
|
||||
|
||||
const connect = () => {
|
||||
this.eventSource = new EventSource(this.endpoint);
|
||||
|
||||
this.eventSource.onmessage = (event) => {
|
||||
this.logData += event.data;
|
||||
this.render();
|
||||
};
|
||||
|
||||
this.eventSource.onerror = (err) => {
|
||||
// Close the current connection
|
||||
this.eventSource.close();
|
||||
|
||||
this.logData += "\nConnection lost. Retrying in 5 seconds...\n";
|
||||
this.render();
|
||||
|
||||
// Attempt to reconnect after 5 seconds
|
||||
setTimeout(() => {
|
||||
this.logData += "Attempting to reconnect...\n";
|
||||
this.render();
|
||||
connect();
|
||||
}, 5000);
|
||||
};
|
||||
};
|
||||
|
||||
// Initial connection
|
||||
connect();
|
||||
}
|
||||
|
||||
render() {
|
||||
let content = this.logData;
|
||||
|
||||
if (this.regexFilter) {
|
||||
const lines = content.split('\n');
|
||||
const filteredLines = lines.filter(line => this.regexFilter.test(line));
|
||||
content = filteredLines.length > 0 ? filteredLines.join('\n') + '\n' : "";
|
||||
}
|
||||
|
||||
this.streamElement.textContent = content;
|
||||
this.streamElement.scrollTop = this.streamElement.scrollHeight;
|
||||
}
|
||||
|
||||
updateFilter() {
|
||||
const pattern = this.filterInput.value.trim();
|
||||
this.filterInput.classList.remove('regex-error');
|
||||
|
||||
if (!pattern) {
|
||||
this.regexFilter = null;
|
||||
this.render();
|
||||
return;
|
||||
}
|
||||
|
||||
function updateFilter() {
|
||||
const pattern = filterInput.value.trim();
|
||||
filterInput.classList.remove('regex-error');
|
||||
if (pattern) {
|
||||
try {
|
||||
regexFilter = new RegExp(pattern);
|
||||
this.regexFilter = new RegExp(pattern);
|
||||
} catch (e) {
|
||||
console.error("Invalid regex pattern:", e);
|
||||
regexFilter = null;
|
||||
filterInput.classList.add('regex-error');
|
||||
return
|
||||
this.regexFilter = null;
|
||||
this.filterInput.classList.add('regex-error');
|
||||
return;
|
||||
}
|
||||
} else {
|
||||
regexFilter = null;
|
||||
}
|
||||
|
||||
render();
|
||||
this.render();
|
||||
}
|
||||
}
|
||||
|
||||
filterInput.addEventListener('input', updateFilter);
|
||||
document.getElementById('clear-button').addEventListener('click', () => {
|
||||
filterInput.value = "";
|
||||
regexFilter = null;
|
||||
render();
|
||||
// Initialize both log streams
|
||||
document.addEventListener('DOMContentLoaded', () => {
|
||||
new LogStream(
|
||||
document.getElementById('proxy-log-stream'),
|
||||
document.getElementById('proxy-filter-input'),
|
||||
document.getElementById('proxy-clear-button'),
|
||||
"/logs/streamSSE/proxy"
|
||||
);
|
||||
|
||||
new LogStream(
|
||||
document.getElementById('upstream-log-stream'),
|
||||
document.getElementById('upstream-filter-input'),
|
||||
document.getElementById('upstream-clear-button'),
|
||||
"/logs/streamSSE/upstream"
|
||||
);
|
||||
|
||||
// Initialize clickable headers
|
||||
document.querySelectorAll('h2').forEach(header => {
|
||||
header.addEventListener('click', () => {
|
||||
const column = header.closest('.log-column');
|
||||
column.classList.toggle('minimized');
|
||||
});
|
||||
});
|
||||
});
|
||||
setupEventSource();
|
||||
updateFilter();
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
@@ -2,11 +2,21 @@ package proxy
|
||||
|
||||
import (
|
||||
"container/ring"
|
||||
"fmt"
|
||||
"io"
|
||||
"os"
|
||||
"sync"
|
||||
)
|
||||
|
||||
type LogLevel int
|
||||
|
||||
const (
|
||||
LevelDebug LogLevel = iota
|
||||
LevelInfo
|
||||
LevelWarn
|
||||
LevelError
|
||||
)
|
||||
|
||||
type LogMonitor struct {
|
||||
clients map[chan []byte]bool
|
||||
mu sync.RWMutex
|
||||
@@ -15,6 +25,10 @@ type LogMonitor struct {
|
||||
|
||||
// typically this can be os.Stdout
|
||||
stdout io.Writer
|
||||
|
||||
// logging levels
|
||||
level LogLevel
|
||||
prefix string
|
||||
}
|
||||
|
||||
func NewLogMonitor() *LogMonitor {
|
||||
@@ -26,6 +40,8 @@ func NewLogMonitorWriter(stdout io.Writer) *LogMonitor {
|
||||
clients: make(map[chan []byte]bool),
|
||||
buffer: ring.New(10 * 1024), // keep 10KB of buffered logs
|
||||
stdout: stdout,
|
||||
level: LevelInfo,
|
||||
prefix: "",
|
||||
}
|
||||
}
|
||||
|
||||
@@ -94,3 +110,77 @@ func (w *LogMonitor) broadcast(msg []byte) {
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
func (w *LogMonitor) SetPrefix(prefix string) {
|
||||
w.mu.Lock()
|
||||
defer w.mu.Unlock()
|
||||
w.prefix = prefix
|
||||
}
|
||||
|
||||
func (w *LogMonitor) SetLogLevel(level LogLevel) {
|
||||
w.mu.Lock()
|
||||
defer w.mu.Unlock()
|
||||
w.level = level
|
||||
}
|
||||
|
||||
func (w *LogMonitor) formatMessage(level string, msg string) []byte {
|
||||
prefix := ""
|
||||
if w.prefix != "" {
|
||||
prefix = fmt.Sprintf("[%s] ", w.prefix)
|
||||
}
|
||||
return []byte(fmt.Sprintf("%s[%s] %s\n", prefix, level, msg))
|
||||
}
|
||||
|
||||
func (w *LogMonitor) log(level LogLevel, msg string) {
|
||||
if level < w.level {
|
||||
return
|
||||
}
|
||||
w.Write(w.formatMessage(level.String(), msg))
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Debug(msg string) {
|
||||
w.log(LevelDebug, msg)
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Info(msg string) {
|
||||
w.log(LevelInfo, msg)
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Warn(msg string) {
|
||||
w.log(LevelWarn, msg)
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Error(msg string) {
|
||||
w.log(LevelError, msg)
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Debugf(format string, args ...interface{}) {
|
||||
w.log(LevelDebug, fmt.Sprintf(format, args...))
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Infof(format string, args ...interface{}) {
|
||||
w.log(LevelInfo, fmt.Sprintf(format, args...))
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Warnf(format string, args ...interface{}) {
|
||||
w.log(LevelWarn, fmt.Sprintf(format, args...))
|
||||
}
|
||||
|
||||
func (w *LogMonitor) Errorf(format string, args ...interface{}) {
|
||||
w.log(LevelError, fmt.Sprintf(format, args...))
|
||||
}
|
||||
|
||||
func (l LogLevel) String() string {
|
||||
switch l {
|
||||
case LevelDebug:
|
||||
return "DEBUG"
|
||||
case LevelInfo:
|
||||
return "INFO"
|
||||
case LevelWarn:
|
||||
return "WARN"
|
||||
case LevelError:
|
||||
return "ERROR"
|
||||
default:
|
||||
return "UNKNOWN"
|
||||
}
|
||||
}
|
||||
|
||||
+125
-126
@@ -30,11 +30,15 @@ const (
|
||||
)
|
||||
|
||||
type Process struct {
|
||||
ID string
|
||||
config ModelConfig
|
||||
cmd *exec.Cmd
|
||||
logMonitor *LogMonitor
|
||||
healthCheckTimeout int
|
||||
ID string
|
||||
config ModelConfig
|
||||
cmd *exec.Cmd
|
||||
|
||||
processLogger *LogMonitor
|
||||
proxyLogger *LogMonitor
|
||||
|
||||
healthCheckTimeout int
|
||||
healthCheckLoopInterval time.Duration
|
||||
|
||||
lastRequestHandled time.Time
|
||||
|
||||
@@ -51,54 +55,68 @@ type Process struct {
|
||||
shutdownCancel context.CancelFunc
|
||||
}
|
||||
|
||||
func NewProcess(ID string, healthCheckTimeout int, config ModelConfig, logMonitor *LogMonitor) *Process {
|
||||
func NewProcess(ID string, healthCheckTimeout int, config ModelConfig, processLogger *LogMonitor, proxyLogger *LogMonitor) *Process {
|
||||
ctx, cancel := context.WithCancel(context.Background())
|
||||
return &Process{
|
||||
ID: ID,
|
||||
config: config,
|
||||
cmd: nil,
|
||||
logMonitor: logMonitor,
|
||||
healthCheckTimeout: healthCheckTimeout,
|
||||
state: StateStopped,
|
||||
shutdownCtx: ctx,
|
||||
shutdownCancel: cancel,
|
||||
ID: ID,
|
||||
config: config,
|
||||
cmd: nil,
|
||||
processLogger: processLogger,
|
||||
proxyLogger: proxyLogger,
|
||||
healthCheckTimeout: healthCheckTimeout,
|
||||
healthCheckLoopInterval: 5 * time.Second, /* default, can not be set by user - used for testing */
|
||||
state: StateStopped,
|
||||
shutdownCtx: ctx,
|
||||
shutdownCancel: cancel,
|
||||
}
|
||||
}
|
||||
|
||||
func (p *Process) setState(newState ProcessState) error {
|
||||
// enforce valid state transitions
|
||||
invalidTransition := false
|
||||
if p.state == StateStopped {
|
||||
// stopped -> starting
|
||||
if newState != StateStarting {
|
||||
invalidTransition = true
|
||||
}
|
||||
} else if p.state == StateStarting {
|
||||
// starting -> ready | failed | stopping
|
||||
if newState != StateReady && newState != StateFailed && newState != StateStopping {
|
||||
invalidTransition = true
|
||||
}
|
||||
} else if p.state == StateReady {
|
||||
// ready -> stopping
|
||||
if newState != StateStopping {
|
||||
invalidTransition = true
|
||||
}
|
||||
} else if p.state == StateStopping {
|
||||
// stopping -> stopped | shutdown
|
||||
if newState != StateStopped && newState != StateShutdown {
|
||||
invalidTransition = true
|
||||
}
|
||||
} else if p.state == StateFailed || p.state == StateShutdown {
|
||||
invalidTransition = true
|
||||
// LogMonitor returns the log monitor associated with the process.
|
||||
func (p *Process) LogMonitor() *LogMonitor {
|
||||
return p.processLogger
|
||||
}
|
||||
|
||||
// custom error types for swapping state
|
||||
var (
|
||||
ErrExpectedStateMismatch = errors.New("expected state mismatch")
|
||||
ErrInvalidStateTransition = errors.New("invalid state transition")
|
||||
)
|
||||
|
||||
// swapState performs a compare and swap of the state atomically. It returns the current state
|
||||
// and an error if the swap failed.
|
||||
func (p *Process) swapState(expectedState, newState ProcessState) (ProcessState, error) {
|
||||
p.stateMutex.Lock()
|
||||
defer p.stateMutex.Unlock()
|
||||
|
||||
if p.state != expectedState {
|
||||
return p.state, ErrExpectedStateMismatch
|
||||
}
|
||||
|
||||
if invalidTransition {
|
||||
//panic(fmt.Sprintf("Invalid state transition from %s to %s", p.state, newState))
|
||||
return fmt.Errorf("invalid state transition from %s to %s", p.state, newState)
|
||||
if !isValidTransition(p.state, newState) {
|
||||
p.proxyLogger.Warnf("Invalid state transition from %s to %s", p.state, newState)
|
||||
return p.state, ErrInvalidStateTransition
|
||||
}
|
||||
|
||||
p.proxyLogger.Debugf("State transition from %s to %s", expectedState, newState)
|
||||
p.state = newState
|
||||
return nil
|
||||
return p.state, nil
|
||||
}
|
||||
|
||||
// Helper function to encapsulate transition rules
|
||||
func isValidTransition(from, to ProcessState) bool {
|
||||
switch from {
|
||||
case StateStopped:
|
||||
return to == StateStarting
|
||||
case StateStarting:
|
||||
return to == StateReady || to == StateFailed || to == StateStopping
|
||||
case StateReady:
|
||||
return to == StateStopping
|
||||
case StateStopping:
|
||||
return to == StateStopped || to == StateShutdown
|
||||
case StateFailed, StateShutdown:
|
||||
return false // No transitions allowed from these states
|
||||
}
|
||||
return false
|
||||
}
|
||||
|
||||
func (p *Process) CurrentState() ProcessState {
|
||||
@@ -116,65 +134,48 @@ func (p *Process) start() error {
|
||||
return fmt.Errorf("can not start(), upstream proxy missing")
|
||||
}
|
||||
|
||||
// multiple start() calls will wait for the one that is actually starting to
|
||||
// complete before proceeding.
|
||||
// ===========
|
||||
curState := p.CurrentState()
|
||||
|
||||
if curState == StateReady {
|
||||
return nil
|
||||
}
|
||||
|
||||
if curState == StateStarting {
|
||||
p.waitStarting.Wait()
|
||||
|
||||
if state := p.CurrentState(); state != StateReady {
|
||||
return fmt.Errorf("start() failed current state: %v", state)
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
// ===========
|
||||
|
||||
// There is the possibility of a hard to replicate race condition where
|
||||
// curState *WAS* StateStopped but by the time we get to the p.stateMutex.Lock()
|
||||
// below, it's value has changed!
|
||||
|
||||
p.stateMutex.Lock()
|
||||
defer p.stateMutex.Unlock()
|
||||
|
||||
// with the exclusive lock, check if p.state is StateStopped, which is the only valid state
|
||||
// to transition from to StateReady
|
||||
|
||||
if p.state != StateStopped {
|
||||
if p.state == StateReady {
|
||||
return nil
|
||||
} else {
|
||||
return fmt.Errorf("start() can not proceed expected StateReady but process is in %v", p.state)
|
||||
}
|
||||
}
|
||||
|
||||
if err := p.setState(StateStarting); err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
p.waitStarting.Add(1)
|
||||
defer p.waitStarting.Done()
|
||||
|
||||
args, err := p.config.SanitizedCommand()
|
||||
if err != nil {
|
||||
return fmt.Errorf("unable to get sanitized command: %v", err)
|
||||
}
|
||||
|
||||
if curState, err := p.swapState(StateStopped, StateStarting); err != nil {
|
||||
if err == ErrExpectedStateMismatch {
|
||||
// already starting, just wait for it to complete and expect
|
||||
// it to be be in the Ready start after. If not, return an error
|
||||
if curState == StateStarting {
|
||||
p.waitStarting.Wait()
|
||||
if state := p.CurrentState(); state == StateReady {
|
||||
return nil
|
||||
} else {
|
||||
return fmt.Errorf("process was already starting but wound up in state %v", state)
|
||||
}
|
||||
} else {
|
||||
return fmt.Errorf("processes was in state %v when start() was called", curState)
|
||||
}
|
||||
} else {
|
||||
return fmt.Errorf("failed to set Process state to starting: current state: %v, error: %v", curState, err)
|
||||
}
|
||||
}
|
||||
|
||||
p.waitStarting.Add(1)
|
||||
defer p.waitStarting.Done()
|
||||
|
||||
p.cmd = exec.Command(args[0], args[1:]...)
|
||||
p.cmd.Stdout = p.logMonitor
|
||||
p.cmd.Stderr = p.logMonitor
|
||||
p.cmd.Stdout = p.processLogger
|
||||
p.cmd.Stderr = p.processLogger
|
||||
p.cmd.Env = p.config.Env
|
||||
|
||||
err = p.cmd.Start()
|
||||
|
||||
// Set process state to failed
|
||||
if err != nil {
|
||||
p.setState(StateFailed)
|
||||
if curState, swapErr := p.swapState(StateStarting, StateFailed); swapErr != nil {
|
||||
return fmt.Errorf(
|
||||
"failed to start command and state swap failed. command error: %v, current state: %v, state swap error: %v",
|
||||
err, curState, swapErr,
|
||||
)
|
||||
}
|
||||
return fmt.Errorf("start() failed: %v", err)
|
||||
}
|
||||
|
||||
@@ -209,31 +210,35 @@ func (p *Process) start() error {
|
||||
)
|
||||
defer cancelHealthCheck()
|
||||
|
||||
// Health check loop
|
||||
loop:
|
||||
// Ready Check loop
|
||||
for {
|
||||
select {
|
||||
case <-checkDeadline.Done():
|
||||
p.setState(StateFailed)
|
||||
return fmt.Errorf("health check failed after %vs", maxDuration.Seconds())
|
||||
if curState, err := p.swapState(StateStarting, StateFailed); err != nil {
|
||||
return fmt.Errorf("health check timed out after %vs AND state swap failed: %v, current state: %v", maxDuration.Seconds(), err, curState)
|
||||
} else {
|
||||
return fmt.Errorf("health check timed out after %vs", maxDuration.Seconds())
|
||||
}
|
||||
case <-p.shutdownCtx.Done():
|
||||
return errors.New("health check interrupted due to shutdown")
|
||||
default:
|
||||
if err := p.checkHealthEndpoint(healthURL); err == nil {
|
||||
p.proxyLogger.Infof("Health check passed on %s", healthURL)
|
||||
cancelHealthCheck()
|
||||
break loop
|
||||
} else {
|
||||
if strings.Contains(err.Error(), "connection refused") {
|
||||
endTime, _ := checkDeadline.Deadline()
|
||||
ttl := time.Until(endTime)
|
||||
fmt.Fprintf(p.logMonitor, "!!! Connection refused on %s, ttl %.0fs\n", healthURL, ttl.Seconds())
|
||||
p.proxyLogger.Infof("Connection refused on %s, retrying in %.0fs", healthURL, ttl.Seconds())
|
||||
} else {
|
||||
fmt.Fprintf(p.logMonitor, "!!! Health check error: %v\n", err)
|
||||
p.proxyLogger.Infof("Health check error on %s, %v", healthURL, err)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
<-time.After(5 * time.Second)
|
||||
<-time.After(p.healthCheckLoopInterval)
|
||||
}
|
||||
}
|
||||
|
||||
@@ -244,7 +249,7 @@ func (p *Process) start() error {
|
||||
maxDuration := time.Duration(p.config.UnloadAfter) * time.Second
|
||||
|
||||
for range time.Tick(time.Second) {
|
||||
if p.state != StateReady {
|
||||
if p.CurrentState() != StateReady {
|
||||
return
|
||||
}
|
||||
|
||||
@@ -252,7 +257,8 @@ func (p *Process) start() error {
|
||||
p.inFlightRequests.Wait()
|
||||
|
||||
if time.Since(p.lastRequestHandled) > maxDuration {
|
||||
fmt.Fprintf(p.logMonitor, "!!! Unloading model %s, TTL of %ds reached.\n", p.ID, p.config.UnloadAfter)
|
||||
|
||||
p.proxyLogger.Infof("Unloading model %s, TTL of %ds reached.", p.ID, p.config.UnloadAfter)
|
||||
p.Stop()
|
||||
return
|
||||
}
|
||||
@@ -260,26 +266,28 @@ func (p *Process) start() error {
|
||||
}()
|
||||
}
|
||||
|
||||
return p.setState(StateReady)
|
||||
if curState, err := p.swapState(StateStarting, StateReady); err != nil {
|
||||
return fmt.Errorf("failed to set Process state to ready: current state: %v, error: %v", curState, err)
|
||||
} else {
|
||||
return nil
|
||||
}
|
||||
}
|
||||
|
||||
func (p *Process) Stop() {
|
||||
// wait for any inflight requests before proceeding
|
||||
p.inFlightRequests.Wait()
|
||||
p.stateMutex.Lock()
|
||||
defer p.stateMutex.Unlock()
|
||||
|
||||
// calling Stop() when state is invalid is a no-op
|
||||
if err := p.setState(StateStopping); err != nil {
|
||||
fmt.Fprintf(p.logMonitor, "!!! Info - Stop() err: %v\n", err)
|
||||
if curState, err := p.swapState(StateReady, StateStopping); err != nil {
|
||||
p.proxyLogger.Infof("Stop() Ready -> StateStopping err: %v, current state: %v", err, curState)
|
||||
return
|
||||
}
|
||||
|
||||
// stop the process with a graceful exit timeout
|
||||
p.stopCommand(5 * time.Second)
|
||||
|
||||
if err := p.setState(StateStopped); err != nil {
|
||||
panic(fmt.Sprintf("Stop() failed to set state to stopped: %v", err))
|
||||
if curState, err := p.swapState(StateStopping, StateStopped); err != nil {
|
||||
p.proxyLogger.Infof("Stop() StateStopping -> StateStopped err: %v, current state: %v", err, curState)
|
||||
}
|
||||
}
|
||||
|
||||
@@ -287,19 +295,9 @@ func (p *Process) Stop() {
|
||||
// of time for any inflight requests to complete before shutting down. If the Process
|
||||
// is in the state of starting, it will cancel it and shut it down
|
||||
func (p *Process) Shutdown() {
|
||||
// cancel anything that can be interrupted by a shutdown (ie: healthcheck)
|
||||
p.shutdownCancel()
|
||||
|
||||
p.stateMutex.Lock()
|
||||
defer p.stateMutex.Unlock()
|
||||
p.setState(StateStopping)
|
||||
|
||||
// 5 seconds to stop the process
|
||||
p.stopCommand(5 * time.Second)
|
||||
if err := p.setState(StateShutdown); err != nil {
|
||||
fmt.Printf("!!! Shutdown() failed to set state to shutdown: %v", err)
|
||||
}
|
||||
p.setState(StateShutdown)
|
||||
p.state = StateShutdown
|
||||
}
|
||||
|
||||
// stopCommand will send a SIGTERM to the process and wait for it to exit.
|
||||
@@ -314,31 +312,32 @@ func (p *Process) stopCommand(sigtermTTL time.Duration) {
|
||||
}()
|
||||
|
||||
if p.cmd == nil || p.cmd.Process == nil {
|
||||
fmt.Fprintf(p.logMonitor, "!!! process [%s] cmd or cmd.Process is nil", p.ID)
|
||||
p.proxyLogger.Warnf("Process [%s] cmd or cmd.Process is nil", p.ID)
|
||||
return
|
||||
}
|
||||
|
||||
p.cmd.Process.Signal(syscall.SIGTERM)
|
||||
if err := p.terminateProcess(); err != nil {
|
||||
p.proxyLogger.Infof("Failed to gracefully terminate process [%s]: %v", p.ID, err)
|
||||
}
|
||||
|
||||
select {
|
||||
case <-sigtermTimeout.Done():
|
||||
fmt.Fprintf(p.logMonitor, "!!! process [%s] timed out waiting to stop, sending KILL signal\n", p.ID)
|
||||
p.proxyLogger.Infof("Process [%s] timed out waiting to stop, sending KILL signal", p.ID)
|
||||
p.cmd.Process.Kill()
|
||||
case err := <-sigtermNormal:
|
||||
if err != nil {
|
||||
if errno, ok := err.(syscall.Errno); ok {
|
||||
fmt.Fprintf(p.logMonitor, "!!! process [%s] errno >> %v\n", p.ID, errno)
|
||||
p.proxyLogger.Errorf("Process [%s] errno >> %v", p.ID, errno)
|
||||
} else if exitError, ok := err.(*exec.ExitError); ok {
|
||||
if strings.Contains(exitError.String(), "signal: terminated") {
|
||||
fmt.Fprintf(p.logMonitor, "!!! process [%s] stopped OK\n", p.ID)
|
||||
p.proxyLogger.Infof("Process [%s] stopped OK", p.ID)
|
||||
} else if strings.Contains(exitError.String(), "signal: interrupt") {
|
||||
fmt.Fprintf(p.logMonitor, "!!! process [%s] interrupted OK\n", p.ID)
|
||||
p.proxyLogger.Infof("Process [%s] interrupted OK", p.ID)
|
||||
} else {
|
||||
fmt.Fprintf(p.logMonitor, "!!! process [%s] ExitError >> %v, exit code: %d\n", p.ID, exitError, exitError.ExitCode())
|
||||
p.proxyLogger.Warnf("Process [%s] ExitError >> %v, exit code: %d", p.ID, exitError, exitError.ExitCode())
|
||||
}
|
||||
|
||||
} else {
|
||||
fmt.Fprintf(p.logMonitor, "!!! process [%s] exited >> %v\n", p.ID, err)
|
||||
p.proxyLogger.Errorf("Process [%s] exited >> %v", p.ID, err)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -0,0 +1,9 @@
|
||||
//go:build !windows
|
||||
|
||||
package proxy
|
||||
|
||||
import "syscall"
|
||||
|
||||
func (p *Process) terminateProcess() error {
|
||||
return p.cmd.Process.Signal(syscall.SIGTERM)
|
||||
}
|
||||
@@ -0,0 +1,14 @@
|
||||
//go:build windows
|
||||
|
||||
package proxy
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"os/exec"
|
||||
)
|
||||
|
||||
func (p *Process) terminateProcess() error {
|
||||
pid := fmt.Sprintf("%d", p.cmd.Process.Pid)
|
||||
cmd := exec.Command("taskkill", "/f", "/t", "/pid", pid)
|
||||
return cmd.Run()
|
||||
}
|
||||
+40
-35
@@ -5,7 +5,6 @@ import (
|
||||
"io"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"os"
|
||||
"sync"
|
||||
"testing"
|
||||
"time"
|
||||
@@ -13,13 +12,17 @@ import (
|
||||
"github.com/stretchr/testify/assert"
|
||||
)
|
||||
|
||||
var (
|
||||
discardLogger = NewLogMonitorWriter(io.Discard)
|
||||
)
|
||||
|
||||
func TestProcess_AutomaticallyStartsUpstream(t *testing.T) {
|
||||
logMonitor := NewLogMonitorWriter(io.Discard)
|
||||
|
||||
expectedMessage := "testing91931"
|
||||
config := getTestSimpleResponderConfig(expectedMessage)
|
||||
|
||||
// Create a process
|
||||
process := NewProcess("test-process", 5, config, logMonitor)
|
||||
process := NewProcess("test-process", 5, config, discardLogger, discardLogger)
|
||||
defer process.Stop()
|
||||
|
||||
req := httptest.NewRequest("GET", "/test", nil)
|
||||
@@ -52,11 +55,10 @@ func TestProcess_AutomaticallyStartsUpstream(t *testing.T) {
|
||||
// are all handled successfully, even though they all may ask for the process to .start()
|
||||
func TestProcess_WaitOnMultipleStarts(t *testing.T) {
|
||||
|
||||
logMonitor := NewLogMonitorWriter(io.Discard)
|
||||
expectedMessage := "testing91931"
|
||||
config := getTestSimpleResponderConfig(expectedMessage)
|
||||
|
||||
process := NewProcess("test-process", 5, config, logMonitor)
|
||||
process := NewProcess("test-process", 5, config, discardLogger, discardLogger)
|
||||
defer process.Stop()
|
||||
|
||||
var wg sync.WaitGroup
|
||||
@@ -84,7 +86,7 @@ func TestProcess_BrokenModelConfig(t *testing.T) {
|
||||
CheckEndpoint: "/health",
|
||||
}
|
||||
|
||||
process := NewProcess("broken", 1, config, NewLogMonitor())
|
||||
process := NewProcess("broken", 1, config, discardLogger, discardLogger)
|
||||
|
||||
req := httptest.NewRequest("GET", "/", nil)
|
||||
w := httptest.NewRecorder()
|
||||
@@ -109,7 +111,7 @@ func TestProcess_UnloadAfterTTL(t *testing.T) {
|
||||
config.UnloadAfter = 3 // seconds
|
||||
assert.Equal(t, 3, config.UnloadAfter)
|
||||
|
||||
process := NewProcess("ttl_test", 2, config, NewLogMonitorWriter(io.Discard))
|
||||
process := NewProcess("ttl_test", 2, config, discardLogger, discardLogger)
|
||||
defer process.Stop()
|
||||
|
||||
// this should take 4 seconds
|
||||
@@ -151,7 +153,7 @@ func TestProcess_LowTTLValue(t *testing.T) {
|
||||
config.UnloadAfter = 1 // second
|
||||
assert.Equal(t, 1, config.UnloadAfter)
|
||||
|
||||
process := NewProcess("ttl", 2, config, NewLogMonitorWriter(os.Stdout))
|
||||
process := NewProcess("ttl", 2, config, discardLogger, discardLogger)
|
||||
defer process.Stop()
|
||||
|
||||
for i := 0; i < 100; i++ {
|
||||
@@ -178,7 +180,7 @@ func TestProcess_HTTPRequestsHaveTimeToFinish(t *testing.T) {
|
||||
|
||||
expectedMessage := "12345"
|
||||
config := getTestSimpleResponderConfig(expectedMessage)
|
||||
process := NewProcess("t", 10, config, NewLogMonitorWriter(os.Stdout))
|
||||
process := NewProcess("t", 10, config, discardLogger, discardLogger)
|
||||
defer process.Stop()
|
||||
|
||||
results := map[string]string{
|
||||
@@ -225,39 +227,40 @@ func TestProcess_HTTPRequestsHaveTimeToFinish(t *testing.T) {
|
||||
}
|
||||
}
|
||||
|
||||
func TestSetState(t *testing.T) {
|
||||
func TestProcess_SwapState(t *testing.T) {
|
||||
tests := []struct {
|
||||
name string
|
||||
currentState ProcessState
|
||||
expectedState ProcessState
|
||||
newState ProcessState
|
||||
expectedError error
|
||||
expectedResult ProcessState
|
||||
}{
|
||||
{"Stopped to Starting", StateStopped, StateStarting, nil, StateStarting},
|
||||
{"Starting to Ready", StateStarting, StateReady, nil, StateReady},
|
||||
{"Starting to Failed", StateStarting, StateFailed, nil, StateFailed},
|
||||
{"Starting to Stopping", StateStarting, StateStopping, nil, StateStopping},
|
||||
{"Ready to Stopping", StateReady, StateStopping, nil, StateStopping},
|
||||
{"Stopping to Stopped", StateStopping, StateStopped, nil, StateStopped},
|
||||
{"Stopping to Shutdown", StateStopping, StateShutdown, nil, StateShutdown},
|
||||
{"Stopped to Ready", StateStopped, StateReady, fmt.Errorf("invalid state transition from stopped to ready"), StateStopped},
|
||||
{"Starting to Stopped", StateStarting, StateStopped, fmt.Errorf("invalid state transition from starting to stopped"), StateStarting},
|
||||
{"Ready to Starting", StateReady, StateStarting, fmt.Errorf("invalid state transition from ready to starting"), StateReady},
|
||||
{"Ready to Failed", StateReady, StateFailed, fmt.Errorf("invalid state transition from ready to failed"), StateReady},
|
||||
{"Stopping to Ready", StateStopping, StateReady, fmt.Errorf("invalid state transition from stopping to ready"), StateStopping},
|
||||
{"Failed to Stopped", StateFailed, StateStopped, fmt.Errorf("invalid state transition from failed to stopped"), StateFailed},
|
||||
{"Failed to Starting", StateFailed, StateStarting, fmt.Errorf("invalid state transition from failed to starting"), StateFailed},
|
||||
{"Shutdown to Stopped", StateShutdown, StateStopped, fmt.Errorf("invalid state transition from shutdown to stopped"), StateShutdown},
|
||||
{"Shutdown to Starting", StateShutdown, StateStarting, fmt.Errorf("invalid state transition from shutdown to starting"), StateShutdown},
|
||||
{"Stopped to Starting", StateStopped, StateStopped, StateStarting, nil, StateStarting},
|
||||
{"Starting to Ready", StateStarting, StateStarting, StateReady, nil, StateReady},
|
||||
{"Starting to Failed", StateStarting, StateStarting, StateFailed, nil, StateFailed},
|
||||
{"Starting to Stopping", StateStarting, StateStarting, StateStopping, nil, StateStopping},
|
||||
{"Ready to Stopping", StateReady, StateReady, StateStopping, nil, StateStopping},
|
||||
{"Stopping to Stopped", StateStopping, StateStopping, StateStopped, nil, StateStopped},
|
||||
{"Stopping to Shutdown", StateStopping, StateStopping, StateShutdown, nil, StateShutdown},
|
||||
{"Stopped to Ready", StateStopped, StateStopped, StateReady, ErrInvalidStateTransition, StateStopped},
|
||||
{"Starting to Stopped", StateStarting, StateStarting, StateStopped, ErrInvalidStateTransition, StateStarting},
|
||||
{"Ready to Starting", StateReady, StateReady, StateStarting, ErrInvalidStateTransition, StateReady},
|
||||
{"Ready to Failed", StateReady, StateReady, StateFailed, ErrInvalidStateTransition, StateReady},
|
||||
{"Stopping to Ready", StateStopping, StateStopping, StateReady, ErrInvalidStateTransition, StateStopping},
|
||||
{"Failed to Stopped", StateFailed, StateFailed, StateStopped, ErrInvalidStateTransition, StateFailed},
|
||||
{"Failed to Starting", StateFailed, StateFailed, StateStarting, ErrInvalidStateTransition, StateFailed},
|
||||
{"Shutdown to Stopped", StateShutdown, StateShutdown, StateStopped, ErrInvalidStateTransition, StateShutdown},
|
||||
{"Shutdown to Starting", StateShutdown, StateShutdown, StateStarting, ErrInvalidStateTransition, StateShutdown},
|
||||
{"Expected state mismatch", StateStopped, StateStarting, StateStarting, ErrExpectedStateMismatch, StateStopped},
|
||||
}
|
||||
|
||||
for _, test := range tests {
|
||||
t.Run(test.name, func(t *testing.T) {
|
||||
p := &Process{
|
||||
state: test.currentState,
|
||||
}
|
||||
p := NewProcess("test", 10, getTestSimpleResponderConfig("test"), discardLogger, discardLogger)
|
||||
p.state = test.currentState
|
||||
|
||||
err := p.setState(test.newState)
|
||||
resultState, err := p.swapState(test.expectedState, test.newState)
|
||||
if err != nil && test.expectedError == nil {
|
||||
t.Errorf("Unexpected error: %v", err)
|
||||
} else if err == nil && test.expectedError != nil {
|
||||
@@ -268,8 +271,8 @@ func TestSetState(t *testing.T) {
|
||||
}
|
||||
}
|
||||
|
||||
if p.state != test.expectedResult {
|
||||
t.Errorf("Expected state: %v, got: %v", test.expectedResult, p.state)
|
||||
if resultState != test.expectedResult {
|
||||
t.Errorf("Expected state: %v, got: %v", test.expectedResult, resultState)
|
||||
}
|
||||
})
|
||||
}
|
||||
@@ -280,7 +283,6 @@ func TestProcess_ShutdownInterruptsHealthCheck(t *testing.T) {
|
||||
t.Skip("skipping long shutdown test")
|
||||
}
|
||||
|
||||
logMonitor := NewLogMonitorWriter(io.Discard)
|
||||
expectedMessage := "testing91931"
|
||||
|
||||
// make a config where the healthcheck will always fail because port is wrong
|
||||
@@ -288,13 +290,16 @@ func TestProcess_ShutdownInterruptsHealthCheck(t *testing.T) {
|
||||
config.Proxy = "http://localhost:9998/test"
|
||||
|
||||
healthCheckTTLSeconds := 30
|
||||
process := NewProcess("test-process", healthCheckTTLSeconds, config, logMonitor)
|
||||
process := NewProcess("test-process", healthCheckTTLSeconds, config, discardLogger, discardLogger)
|
||||
|
||||
// make it a lot faster
|
||||
process.healthCheckLoopInterval = time.Second
|
||||
|
||||
// start a goroutine to simulate a shutdown
|
||||
var wg sync.WaitGroup
|
||||
go func() {
|
||||
defer wg.Done()
|
||||
<-time.After(time.Second * 2)
|
||||
<-time.After(time.Millisecond * 500)
|
||||
process.Shutdown()
|
||||
}()
|
||||
wg.Add(1)
|
||||
|
||||
+77
-37
@@ -7,6 +7,7 @@ import (
|
||||
"io"
|
||||
"mime/multipart"
|
||||
"net/http"
|
||||
"os"
|
||||
"sort"
|
||||
"strconv"
|
||||
"strings"
|
||||
@@ -27,59 +28,96 @@ type ProxyManager struct {
|
||||
|
||||
config *Config
|
||||
currentProcesses map[string]*Process
|
||||
logMonitor *LogMonitor
|
||||
ginEngine *gin.Engine
|
||||
|
||||
// logging
|
||||
proxyLogger *LogMonitor
|
||||
upstreamLogger *LogMonitor
|
||||
muxLogger *LogMonitor
|
||||
}
|
||||
|
||||
func New(config *Config) *ProxyManager {
|
||||
// set up loggers
|
||||
stdoutLogger := NewLogMonitorWriter(os.Stdout)
|
||||
upstreamLogger := NewLogMonitorWriter(stdoutLogger)
|
||||
proxyLogger := NewLogMonitorWriter(stdoutLogger)
|
||||
|
||||
if config.LogRequests {
|
||||
proxyLogger.Warn("LogRequests configuration is deprecated. Use logLevel instead.")
|
||||
}
|
||||
|
||||
switch strings.ToLower(strings.TrimSpace(config.LogLevel)) {
|
||||
case "debug":
|
||||
proxyLogger.SetLogLevel(LevelDebug)
|
||||
case "info":
|
||||
proxyLogger.SetLogLevel(LevelInfo)
|
||||
case "warn":
|
||||
proxyLogger.SetLogLevel(LevelWarn)
|
||||
case "error":
|
||||
proxyLogger.SetLogLevel(LevelError)
|
||||
default:
|
||||
proxyLogger.SetLogLevel(LevelInfo)
|
||||
}
|
||||
|
||||
pm := &ProxyManager{
|
||||
config: config,
|
||||
currentProcesses: make(map[string]*Process),
|
||||
logMonitor: NewLogMonitor(),
|
||||
ginEngine: gin.New(),
|
||||
|
||||
proxyLogger: proxyLogger,
|
||||
muxLogger: stdoutLogger,
|
||||
upstreamLogger: upstreamLogger,
|
||||
}
|
||||
|
||||
if config.LogRequests {
|
||||
pm.ginEngine.Use(func(c *gin.Context) {
|
||||
// Start timer
|
||||
start := time.Now()
|
||||
pm.ginEngine.Use(func(c *gin.Context) {
|
||||
// Start timer
|
||||
start := time.Now()
|
||||
|
||||
// capture these because /upstream/:model rewrites them in c.Next()
|
||||
clientIP := c.ClientIP()
|
||||
method := c.Request.Method
|
||||
path := c.Request.URL.Path
|
||||
// capture these because /upstream/:model rewrites them in c.Next()
|
||||
clientIP := c.ClientIP()
|
||||
method := c.Request.Method
|
||||
path := c.Request.URL.Path
|
||||
|
||||
// Process request
|
||||
c.Next()
|
||||
// Process request
|
||||
c.Next()
|
||||
|
||||
// Stop timer
|
||||
duration := time.Since(start)
|
||||
// Stop timer
|
||||
duration := time.Since(start)
|
||||
|
||||
statusCode := c.Writer.Status()
|
||||
bodySize := c.Writer.Size()
|
||||
statusCode := c.Writer.Status()
|
||||
bodySize := c.Writer.Size()
|
||||
|
||||
fmt.Fprintf(pm.logMonitor, "[llama-swap] %s [%s] \"%s %s %s\" %d %d \"%s\" %v\n",
|
||||
clientIP,
|
||||
time.Now().Format("2006-01-02 15:04:05"),
|
||||
method,
|
||||
path,
|
||||
c.Request.Proto,
|
||||
statusCode,
|
||||
bodySize,
|
||||
c.Request.UserAgent(),
|
||||
duration,
|
||||
)
|
||||
})
|
||||
}
|
||||
pm.proxyLogger.Infof("Request %s \"%s %s %s\" %d %d \"%s\" %v",
|
||||
clientIP,
|
||||
method,
|
||||
path,
|
||||
c.Request.Proto,
|
||||
statusCode,
|
||||
bodySize,
|
||||
c.Request.UserAgent(),
|
||||
duration,
|
||||
)
|
||||
})
|
||||
|
||||
// see: https://github.com/mostlygeek/llama-swap/issues/42
|
||||
// see: issue: #81, #77 and #42 for CORS issues
|
||||
// respond with permissive OPTIONS for any endpoint
|
||||
pm.ginEngine.Use(func(c *gin.Context) {
|
||||
if c.Request.Method == "OPTIONS" {
|
||||
c.Header("Access-Control-Allow-Origin", "*")
|
||||
c.Header("Access-Control-Allow-Methods", "GET, POST, OPTIONS")
|
||||
c.Header("Access-Control-Allow-Headers", "Content-Type, Authorization")
|
||||
c.AbortWithStatus(204)
|
||||
c.Header("Access-Control-Allow-Methods", "GET, POST, PUT, PATCH, DELETE, OPTIONS")
|
||||
|
||||
// allow whatever the client requested by default
|
||||
if headers := c.Request.Header.Get("Access-Control-Request-Headers"); headers != "" {
|
||||
sanitized := SanitizeAccessControlRequestHeaderValues(headers)
|
||||
c.Header("Access-Control-Allow-Headers", sanitized)
|
||||
} else {
|
||||
c.Header(
|
||||
"Access-Control-Allow-Headers",
|
||||
"Content-Type, Authorization, Accept, X-Requested-With",
|
||||
)
|
||||
}
|
||||
c.Header("Access-Control-Max-Age", "86400")
|
||||
c.AbortWithStatus(http.StatusNoContent)
|
||||
return
|
||||
}
|
||||
c.Next()
|
||||
@@ -104,6 +142,8 @@ func New(config *Config) *ProxyManager {
|
||||
pm.ginEngine.GET("/logs", pm.sendLogsHandlers)
|
||||
pm.ginEngine.GET("/logs/stream", pm.streamLogsHandler)
|
||||
pm.ginEngine.GET("/logs/streamSSE", pm.streamLogsHandlerSSE)
|
||||
pm.ginEngine.GET("/logs/stream/:logMonitorID", pm.streamLogsHandler)
|
||||
pm.ginEngine.GET("/logs/streamSSE/:logMonitorID", pm.streamLogsHandlerSSE)
|
||||
|
||||
pm.ginEngine.GET("/upstream", pm.upstreamIndex)
|
||||
pm.ginEngine.Any("/upstream/:model_id/*upstreamPath", pm.proxyToUpstream)
|
||||
@@ -263,19 +303,20 @@ func (pm *ProxyManager) swapModel(requestedModel string) (*Process, error) {
|
||||
requestedProcessKey := ProcessKeyName(profileName, realModelName)
|
||||
|
||||
if process, found := pm.currentProcesses[requestedProcessKey]; found {
|
||||
pm.proxyLogger.Debugf("No-swap, using existing process for model [%s]", requestedModel)
|
||||
return process, nil
|
||||
}
|
||||
|
||||
// stop all running models
|
||||
pm.proxyLogger.Infof("Swapping model to [%s]", requestedModel)
|
||||
pm.stopProcesses()
|
||||
|
||||
if profileName == "" {
|
||||
modelConfig, modelID, found := pm.config.FindConfig(realModelName)
|
||||
if !found {
|
||||
return nil, fmt.Errorf("could not find configuration for %s", realModelName)
|
||||
}
|
||||
|
||||
process := NewProcess(modelID, pm.config.HealthCheckTimeout, modelConfig, pm.logMonitor)
|
||||
process := NewProcess(modelID, pm.config.HealthCheckTimeout, modelConfig, pm.upstreamLogger, pm.proxyLogger)
|
||||
processKey := ProcessKeyName(profileName, modelID)
|
||||
pm.currentProcesses[processKey] = process
|
||||
} else {
|
||||
@@ -286,7 +327,7 @@ func (pm *ProxyManager) swapModel(requestedModel string) (*Process, error) {
|
||||
return nil, fmt.Errorf("could not find configuration for %s in group %s", realModelName, profileName)
|
||||
}
|
||||
|
||||
process := NewProcess(modelID, pm.config.HealthCheckTimeout, modelConfig, pm.logMonitor)
|
||||
process := NewProcess(modelID, pm.config.HealthCheckTimeout, modelConfig, pm.upstreamLogger, pm.proxyLogger)
|
||||
processKey := ProcessKeyName(profileName, modelID)
|
||||
pm.currentProcesses[processKey] = process
|
||||
}
|
||||
@@ -374,7 +415,6 @@ func (pm *ProxyManager) proxyOAIHandler(c *gin.Context) {
|
||||
return
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
c.Request.Body = io.NopCloser(bytes.NewBuffer(bodyBytes))
|
||||
|
||||
@@ -9,7 +9,6 @@ import (
|
||||
)
|
||||
|
||||
func (pm *ProxyManager) sendLogsHandlers(c *gin.Context) {
|
||||
|
||||
accept := c.GetHeader("Accept")
|
||||
if strings.Contains(accept, "text/html") {
|
||||
// Set the Content-Type header to text/html
|
||||
@@ -28,7 +27,7 @@ func (pm *ProxyManager) sendLogsHandlers(c *gin.Context) {
|
||||
}
|
||||
} else {
|
||||
c.Header("Content-Type", "text/plain")
|
||||
history := pm.logMonitor.GetHistory()
|
||||
history := pm.muxLogger.GetHistory()
|
||||
_, err := c.Writer.Write(history)
|
||||
if err != nil {
|
||||
c.AbortWithError(http.StatusInternalServerError, err)
|
||||
@@ -42,8 +41,14 @@ func (pm *ProxyManager) streamLogsHandler(c *gin.Context) {
|
||||
c.Header("Transfer-Encoding", "chunked")
|
||||
c.Header("X-Content-Type-Options", "nosniff")
|
||||
|
||||
ch := pm.logMonitor.Subscribe()
|
||||
defer pm.logMonitor.Unsubscribe(ch)
|
||||
logMonitorId := c.Param("logMonitorID")
|
||||
logger, err := pm.getLogger(logMonitorId)
|
||||
if err != nil {
|
||||
c.String(http.StatusBadRequest, err.Error())
|
||||
return
|
||||
}
|
||||
ch := logger.Subscribe()
|
||||
defer logger.Unsubscribe(ch)
|
||||
|
||||
notify := c.Request.Context().Done()
|
||||
flusher, ok := c.Writer.(http.Flusher)
|
||||
@@ -56,7 +61,7 @@ func (pm *ProxyManager) streamLogsHandler(c *gin.Context) {
|
||||
// Send history first if not skipped
|
||||
|
||||
if !skipHistory {
|
||||
history := pm.logMonitor.GetHistory()
|
||||
history := logger.GetHistory()
|
||||
if len(history) != 0 {
|
||||
c.Writer.Write(history)
|
||||
flusher.Flush()
|
||||
@@ -85,15 +90,21 @@ func (pm *ProxyManager) streamLogsHandlerSSE(c *gin.Context) {
|
||||
c.Header("Connection", "keep-alive")
|
||||
c.Header("X-Content-Type-Options", "nosniff")
|
||||
|
||||
ch := pm.logMonitor.Subscribe()
|
||||
defer pm.logMonitor.Unsubscribe(ch)
|
||||
logMonitorId := c.Param("logMonitorID")
|
||||
logger, err := pm.getLogger(logMonitorId)
|
||||
if err != nil {
|
||||
c.String(http.StatusBadRequest, err.Error())
|
||||
return
|
||||
}
|
||||
ch := logger.Subscribe()
|
||||
defer logger.Unsubscribe(ch)
|
||||
|
||||
notify := c.Request.Context().Done()
|
||||
|
||||
// Send history first if not skipped
|
||||
_, skipHistory := c.GetQuery("no-history")
|
||||
if !skipHistory {
|
||||
history := pm.logMonitor.GetHistory()
|
||||
history := logger.GetHistory()
|
||||
if len(history) != 0 {
|
||||
c.SSEvent("message", string(history))
|
||||
c.Writer.Flush()
|
||||
@@ -111,3 +122,21 @@ func (pm *ProxyManager) streamLogsHandlerSSE(c *gin.Context) {
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// getLogger searches for the appropriate logger based on the logMonitorId
|
||||
func (pm *ProxyManager) getLogger(logMonitorId string) (*LogMonitor, error) {
|
||||
var logger *LogMonitor
|
||||
|
||||
if logMonitorId == "" {
|
||||
// maintain the default
|
||||
logger = pm.muxLogger
|
||||
} else if logMonitorId == "proxy" {
|
||||
logger = pm.proxyLogger
|
||||
} else if logMonitorId == "upstream" {
|
||||
logger = pm.upstreamLogger
|
||||
} else {
|
||||
return nil, fmt.Errorf("invalid logger. Use 'proxy' or 'upstream'")
|
||||
}
|
||||
|
||||
return logger, nil
|
||||
}
|
||||
|
||||
@@ -639,5 +639,72 @@ func TestProxyManager_UseModelName(t *testing.T) {
|
||||
assert.Equal(t, upstreamModelName, response["model"])
|
||||
})
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
func TestProxyManager_CORSOptionsHandler(t *testing.T) {
|
||||
config := &Config{
|
||||
HealthCheckTimeout: 15,
|
||||
Models: map[string]ModelConfig{
|
||||
"model1": getTestSimpleResponderConfig("model1"),
|
||||
},
|
||||
LogRequests: true,
|
||||
}
|
||||
|
||||
tests := []struct {
|
||||
name string
|
||||
method string
|
||||
requestHeaders map[string]string
|
||||
expectedStatus int
|
||||
expectedHeaders map[string]string
|
||||
}{
|
||||
{
|
||||
name: "OPTIONS with no headers",
|
||||
method: "OPTIONS",
|
||||
expectedStatus: http.StatusNoContent,
|
||||
expectedHeaders: map[string]string{
|
||||
"Access-Control-Allow-Origin": "*",
|
||||
"Access-Control-Allow-Methods": "GET, POST, PUT, PATCH, DELETE, OPTIONS",
|
||||
"Access-Control-Allow-Headers": "Content-Type, Authorization, Accept, X-Requested-With",
|
||||
},
|
||||
},
|
||||
{
|
||||
name: "OPTIONS with specific headers",
|
||||
method: "OPTIONS",
|
||||
requestHeaders: map[string]string{
|
||||
"Access-Control-Request-Headers": "X-Custom-Header, Some-Other-Header",
|
||||
},
|
||||
expectedStatus: http.StatusNoContent,
|
||||
expectedHeaders: map[string]string{
|
||||
"Access-Control-Allow-Origin": "*",
|
||||
"Access-Control-Allow-Methods": "GET, POST, PUT, PATCH, DELETE, OPTIONS",
|
||||
"Access-Control-Allow-Headers": "X-Custom-Header, Some-Other-Header",
|
||||
},
|
||||
},
|
||||
{
|
||||
name: "Non-OPTIONS request",
|
||||
method: "GET",
|
||||
expectedStatus: http.StatusNotFound, // Since we don't have a GET route defined
|
||||
},
|
||||
}
|
||||
|
||||
for _, tt := range tests {
|
||||
t.Run(tt.name, func(t *testing.T) {
|
||||
proxy := New(config)
|
||||
defer proxy.StopProcesses()
|
||||
|
||||
req := httptest.NewRequest(tt.method, "/v1/chat/completions", nil)
|
||||
for k, v := range tt.requestHeaders {
|
||||
req.Header.Set(k, v)
|
||||
}
|
||||
|
||||
w := httptest.NewRecorder()
|
||||
proxy.ginEngine.ServeHTTP(w, req)
|
||||
|
||||
assert.Equal(t, tt.expectedStatus, w.Code)
|
||||
|
||||
for header, expectedValue := range tt.expectedHeaders {
|
||||
assert.Equal(t, expectedValue, w.Header().Get(header))
|
||||
}
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
@@ -0,0 +1,43 @@
|
||||
package proxy
|
||||
|
||||
import (
|
||||
"strings"
|
||||
)
|
||||
|
||||
func isTokenChar(r rune) bool {
|
||||
switch {
|
||||
case r >= 'a' && r <= 'z':
|
||||
case r >= 'A' && r <= 'Z':
|
||||
case r >= '0' && r <= '9':
|
||||
case strings.ContainsRune("!#$%&'*+-.^_`|~", r):
|
||||
default:
|
||||
return false
|
||||
}
|
||||
return true
|
||||
}
|
||||
|
||||
func SanitizeAccessControlRequestHeaderValues(headerValues string) string {
|
||||
parts := strings.Split(headerValues, ",")
|
||||
valid := make([]string, 0, len(parts))
|
||||
|
||||
for _, p := range parts {
|
||||
v := strings.TrimSpace(p)
|
||||
if v == "" {
|
||||
continue
|
||||
}
|
||||
|
||||
validPart := true
|
||||
for _, c := range v {
|
||||
if !isTokenChar(c) {
|
||||
validPart = false
|
||||
break
|
||||
}
|
||||
}
|
||||
|
||||
if validPart {
|
||||
valid = append(valid, v)
|
||||
}
|
||||
}
|
||||
|
||||
return strings.Join(valid, ", ")
|
||||
}
|
||||
@@ -0,0 +1,77 @@
|
||||
package proxy
|
||||
|
||||
import "testing"
|
||||
|
||||
func TestSanitizeAccessControlRequestHeaderValues(t *testing.T) {
|
||||
tests := []struct {
|
||||
name string
|
||||
input string
|
||||
expected string
|
||||
}{
|
||||
{
|
||||
name: "empty string",
|
||||
input: "",
|
||||
expected: "",
|
||||
},
|
||||
{
|
||||
name: "whitespace only",
|
||||
input: " ",
|
||||
expected: "",
|
||||
},
|
||||
{
|
||||
name: "single valid value",
|
||||
input: "content-type",
|
||||
expected: "content-type",
|
||||
},
|
||||
{
|
||||
name: "multiple valid values",
|
||||
input: "content-type, authorization, x-requested-with",
|
||||
expected: "content-type, authorization, x-requested-with",
|
||||
},
|
||||
{
|
||||
name: "values with extra spaces",
|
||||
input: " content-type , authorization ",
|
||||
expected: "content-type, authorization",
|
||||
},
|
||||
{
|
||||
name: "values with tabs",
|
||||
input: "content-type,\tauthorization",
|
||||
expected: "content-type, authorization",
|
||||
},
|
||||
{
|
||||
name: "values with invalid characters",
|
||||
input: "content-type, auth\n, x-requested-with\r",
|
||||
expected: "content-type, auth, x-requested-with",
|
||||
},
|
||||
{
|
||||
name: "empty values in list",
|
||||
input: "content-type,,authorization",
|
||||
expected: "content-type, authorization",
|
||||
},
|
||||
{
|
||||
name: "leading and trailing commas",
|
||||
input: ",content-type,authorization,",
|
||||
expected: "content-type, authorization",
|
||||
},
|
||||
{
|
||||
name: "mixed valid and invalid values",
|
||||
input: "content-type, \x00invalid, x-requested-with",
|
||||
expected: "content-type, x-requested-with",
|
||||
},
|
||||
{
|
||||
name: "mixed case values",
|
||||
input: "Content-Type, my-Valid-Header, Another-hEader",
|
||||
expected: "Content-Type, my-Valid-Header, Another-hEader",
|
||||
},
|
||||
}
|
||||
|
||||
for _, tt := range tests {
|
||||
t.Run(tt.name, func(t *testing.T) {
|
||||
got := SanitizeAccessControlRequestHeaderValues(tt.input)
|
||||
if got != tt.expected {
|
||||
t.Errorf("SanitizeAccessControlRequestHeaderValues(%q) = %q, want %q",
|
||||
tt.input, got, tt.expected)
|
||||
}
|
||||
})
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user