feat: add /logs/stream SSE endpoint for real-time log streaming

HF's runtime log API doesn't capture Docker SDK container output
(confirmed: stdout, stderr, /proc/1/fd/*, Python PID 1 all fail).
Added our own SSE streamer at /logs/stream that mimics HF's format:

curl -N https://tao-shen-huggingrun.hf.space/logs/stream

Also available as static file: curl .../logs

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Dockerfile

@@ -47,6 +47,7 @@ RUN mkdir -p /data
 COPY ubuntu-server/nginx.conf /etc/nginx/nginx.conf
 COPY ubuntu-server/ws-ssh-bridge.py /opt/ws-ssh-bridge.py
 COPY ubuntu-server/git_sync_daemon.py /opt/git_sync_daemon.py
+COPY ubuntu-server/log_streamer.py /opt/log_streamer.py
 COPY ubuntu-server/start-server.sh /opt/start-server.sh
 COPY scripts /scripts
 RUN chmod +x /scripts/entrypoint.sh /scripts/entrypoint_wrapper.py /opt/start-server.sh
@@ -58,6 +59,4 @@ ENV SSH_PORT=2222
 
 # Run as root (needed for: apt install persistence, bind mounts, sshd)
 EXPOSE 7860
-# Python PID 1: HF log API captures Python stdout/stderr reliably
-# The wrapper tails /var/log/huggingrun.log → stdout/stderr for HF to capture
-ENTRYPOINT ["python3", "/scripts/entrypoint_wrapper.py"]
+ENTRYPOINT ["/scripts/entrypoint.sh"]

scripts/entrypoint_wrapper.py

@@ -1,84 +0,0 @@
-#!/usr/bin/env python3
-"""
-Python PID 1 wrapper for HF Spaces log compatibility.
-HF's runtime log API captures Python stdout/stderr better than bash.
-This wrapper stays as PID 1, runs entrypoint.sh as a child,
-and mirrors all log output to stdout/stderr.
-"""
-import subprocess
-import sys
-import os
-import signal
-import time
-import threading
-
-LOGFILE = "/var/log/huggingrun.log"
-
-def log(msg):
-    """Write to stdout, stderr, and logfile simultaneously."""
-    line = f"{msg}\n"
-    sys.stdout.write(line)
-    sys.stdout.flush()
-    sys.stderr.write(line)
-    sys.stderr.flush()
-    try:
-        with open(LOGFILE, "a") as f:
-            f.write(line)
-    except Exception:
-        pass
-
-def tail_logfile():
-    """Background thread: tail the logfile and print new lines to stdout/stderr."""
-    # Wait for logfile to exist
-    while not os.path.exists(LOGFILE):
-        time.sleep(0.5)
-
-    with open(LOGFILE, "r") as f:
-        # Skip to current end
-        f.seek(0, 2)
-        while True:
-            line = f.readline()
-            if line:
-                line = line.rstrip("\n")
-                sys.stdout.write(f"{line}\n")
-                sys.stdout.flush()
-                sys.stderr.write(f"{line}\n")
-                sys.stderr.flush()
-            else:
-                time.sleep(0.3)
-
-if __name__ == "__main__":
-    log("========================================")
-    log("[wrapper] HuggingRun Python PID 1 wrapper")
-    log(f"[wrapper] PID={os.getpid()}, Date={time.strftime('%Y-%m-%d %H:%M:%S UTC', time.gmtime())}")
-    log(f"[wrapper] Python {sys.version.split()[0]}")
-    log("========================================")
-
-    # Start background tailer: reads logfile and echos to stdout/stderr
-    # This ensures all log output (from bash scripts, python daemons) gets
-    # printed through PID 1's stdout/stderr for HF to capture
-    tailer = threading.Thread(target=tail_logfile, daemon=True)
-    tailer.start()
-
-    # Run entrypoint.sh as child process
-    proc = subprocess.Popen(
-        ["/bin/bash", "/scripts/entrypoint.sh"],
-        stdout=subprocess.DEVNULL,  # entrypoint writes to logfile
-        stderr=subprocess.DEVNULL,  # entrypoint writes to logfile
-    )
-
-    # Forward signals to child
-    def forward_signal(sig, frame):
-        log(f"[wrapper] Signal {sig} received, forwarding to child PID={proc.pid}")
-        try:
-            proc.send_signal(sig)
-        except ProcessLookupError:
-            pass
-
-    signal.signal(signal.SIGTERM, forward_signal)
-    signal.signal(signal.SIGINT, forward_signal)
-
-    # Wait for child to exit
-    rc = proc.wait()
-    log(f"[wrapper] Child exited with code {rc}")
-    sys.exit(rc)

ubuntu-server/log_streamer.py

@@ -0,0 +1,74 @@
+#!/usr/bin/env python3
+"""
+SSE log streamer — serves /var/log/huggingrun.log as Server-Sent Events.
+Mimics HF's runtime log API format so you can use the same curl -N pattern.
+
+Usage:
+  curl -N https://tao-shen-huggingrun.hf.space/logs/stream
+"""
+import http.server
+import time
+import os
+import json
+from datetime import datetime, timezone
+
+LOGFILE = "/var/log/huggingrun.log"
+PORT = 7863
+
+
+class SSEHandler(http.server.BaseHTTPRequestHandler):
+    def log_message(self, format, *args):
+        pass  # suppress access logs
+
+    def do_GET(self):
+        if self.path == "/stream":
+            self.send_response(200)
+            self.send_header("Content-Type", "text/event-stream")
+            self.send_header("Cache-Control", "no-cache")
+            self.send_header("Connection", "keep-alive")
+            self.send_header("Access-Control-Allow-Origin", "*")
+            self.end_headers()
+
+            try:
+                # Send existing log content first (history)
+                if os.path.exists(LOGFILE):
+                    with open(LOGFILE, "r") as f:
+                        for line in f:
+                            line = line.rstrip("\n")
+                            if line:
+                                event = json.dumps({
+                                    "data": line + "\n",
+                                    "timestamp": datetime.now(timezone.utc).isoformat()
+                                })
+                                self.wfile.write(f"data: {event}\n\n".encode())
+                                self.wfile.flush()
+
+                # Then tail for new lines
+                with open(LOGFILE, "r") as f:
+                    f.seek(0, 2)  # end of file
+                    while True:
+                        line = f.readline()
+                        if line:
+                            line = line.rstrip("\n")
+                            if line:
+                                event = json.dumps({
+                                    "data": line + "\n",
+                                    "timestamp": datetime.now(timezone.utc).isoformat()
+                                })
+                                self.wfile.write(f"data: {event}\n\n".encode())
+                                self.wfile.flush()
+                        else:
+                            # Send keep-alive comment every 15s
+                            self.wfile.write(b": keep-alive\n\n")
+                            self.wfile.flush()
+                            time.sleep(1)
+            except (BrokenPipeError, ConnectionResetError):
+                pass
+        else:
+            self.send_response(404)
+            self.end_headers()
+
+
+if __name__ == "__main__":
+    server = http.server.HTTPServer(("127.0.0.1", PORT), SSEHandler)
+    server.serve_forever()

ubuntu-server/nginx.conf

@@ -33,13 +33,24 @@ http {
             proxy_send_timeout 86400;
         }
 
-        # /logs → live container logs
-        location /logs {
+        # /logs → full log file (static)
+        location = /logs {
             default_type text/plain;
             add_header Cache-Control "no-cache, no-store";
             alias /var/log/huggingrun.log;
         }
 
+        # /logs/stream → SSE real-time log stream (like HF's logs/run API)
+        location /logs/stream {
+            proxy_pass http://127.0.0.1:7863/stream;
+            proxy_http_version 1.1;
+            proxy_set_header Connection "";
+            proxy_buffering off;
+            proxy_cache off;
+            chunked_transfer_encoding off;
+            proxy_read_timeout 86400;
+        }
+
         # Everything else → ttyd web terminal (on 7681)
         location / {
             proxy_pass http://127.0.0.1:7681;

ubuntu-server/start-server.sh

@@ -59,6 +59,17 @@ else
     log "[ubuntu] [FAILED] sshd failed to start"
 fi
 
+# ── SSE log streamer ──────────────────────────────────────────────
+log "[ubuntu] Starting SSE log streamer on 127.0.0.1:7863 ..."
+python3 /opt/log_streamer.py &
+STREAMER_PID=$!
+sleep 1
+if kill -0 $STREAMER_PID 2>/dev/null; then
+    log "[ubuntu] [  OK  ] SSE log streamer started (PID=${STREAMER_PID})"
+else
+    log "[ubuntu] [FAILED] SSE log streamer failed to start"
+fi
+
 # ── WebSocket-to-SSH bridge ───────────────────────────────────────
 log "[ubuntu] Starting WS-SSH bridge on 127.0.0.1:7862 ..."
 python3 /opt/ws-ssh-bridge.py &
@@ -84,10 +95,11 @@ fi
 # ── Process summary ───────────────────────────────────────────────
 log "========================================"
 log "[ubuntu] Services:"
-log "[ubuntu]   nginx         PID=${NGINX_PID}   0.0.0.0:7860"
-log "[ubuntu]   sshd          PID=${SSHD_PID}   127.0.0.1:${SSH_PORT}"
-log "[ubuntu]   ws-ssh-bridge PID=${BRIDGE_PID}  127.0.0.1:7862"
-log "[ubuntu]   ttyd          PID=${TTYD_PID}    127.0.0.1:${TTYD_PORT}"
+log "[ubuntu]   nginx         PID=${NGINX_PID}     0.0.0.0:7860"
+log "[ubuntu]   sshd          PID=${SSHD_PID}     127.0.0.1:${SSH_PORT}"
+log "[ubuntu]   log-streamer  PID=${STREAMER_PID}  127.0.0.1:7863"
+log "[ubuntu]   ws-ssh-bridge PID=${BRIDGE_PID}    127.0.0.1:7862"
+log "[ubuntu]   ttyd          PID=${TTYD_PID}      127.0.0.1:${TTYD_PORT}"
 log "========================================"
 log "[ubuntu] Base packages: $(wc -l < /etc/base-packages.list 2>/dev/null || echo '?')"
 log "[ubuntu] Current packages: $(dpkg-query -W -f='\n' 2>/dev/null | wc -l)"
@@ -96,7 +108,8 @@ log "[ubuntu] All processes:"
 ps aux --no-headers 2>/dev/null | awk '{printf "[ubuntu]   %-8s PID=%-6s %s\n", $1, $2, $11}' | while IFS= read -r line; do log "$line"; done
 
 log "[ubuntu] ══ System ready ══"
-log "[ubuntu] View logs: curl https://<space>.hf.space/logs"
+log "[ubuntu] View logs:  curl https://<space>.hf.space/logs"
+log "[ubuntu] Stream SSE: curl -N https://<space>.hf.space/logs/stream"
 
 # ── Heartbeat ─────────────────────────────────────────────────────
 (while true; do