Merge pull request #1014 from minrk/flush-buffers

repo2docker/buildpacks/base.py

@@ -9,9 +9,6 @@ import string
 import sys
 import hashlib
 import escapism
-import xml.etree.ElementTree as ET
-
-from traitlets import Dict
 
 # Only use syntax features supported by Docker 17.09
 TEMPLATE = r"""
@@ -181,6 +178,8 @@ ENV R2D_ENTRYPOINT "{{ start_script }}"
 {% endif -%}
 
 # Add entrypoint
+ENV PYTHONUNBUFFERED=1
+COPY /python3-login /usr/local/bin/python3-login
 COPY /repo2docker-entrypoint /usr/local/bin/repo2docker-entrypoint
 ENTRYPOINT ["/usr/local/bin/repo2docker-entrypoint"]
 
@@ -193,9 +192,7 @@ CMD ["jupyter", "notebook", "--ip", "0.0.0.0"]
 {% endif %}
 """
 
-ENTRYPOINT_FILE = os.path.join(
-    os.path.dirname(os.path.abspath(__file__)), "repo2docker-entrypoint"
-)
+HERE = os.path.dirname(os.path.abspath(__file__))
 
 # Also used for the group
 DEFAULT_NB_UID = 1000
@@ -582,7 +579,8 @@ class BuildPack:
             dest_path, src_path = self.generate_build_context_filename(src)
             tar.add(src_path, dest_path, filter=_filter_tar)
 
-        tar.add(ENTRYPOINT_FILE, "repo2docker-entrypoint", filter=_filter_tar)
+        for fname in ("repo2docker-entrypoint", "python3-login"):
+            tar.add(os.path.join(HERE, fname), fname, filter=_filter_tar)
 
         tar.add(".", "src/", filter=_filter_tar)
 

repo2docker/buildpacks/python3-login

@@ -0,0 +1,11 @@
+#!/bin/bash -l
+# This is an executable that launches Python in a login shell
+# to ensure that full profile setup occurs.
+# shebang on linux only allows 1 argument,
+# so we couldn't pick a login shell in one shebang line
+# for a Python script
+
+# -u means unbuffered, which one ~always wants in a container
+# otherwise output can be mysteriously missing
+
+exec python3 -u "$@"

repo2docker/buildpacks/repo2docker-entrypoint

@@ -1,24 +1,97 @@
-#!/bin/bash -l
-# lightest possible entrypoint that ensures that
-# we use a login shell to get a fully configured shell environment
-# (e.g. sourcing /etc/profile.d, ~/.bashrc, and friends)
+#!/usr/local/bin/python3-login
+# note: must run on Python >= 3.5, which mainly means no f-strings
 
-# Setup a file descriptor (FD) that is connected to a tee process which
-# writes its input to $REPO_DIR/.jupyter-server-log.txt
-# We later use this FD as a place to redirect the output of the actual
-# command to. We can't add `tee` to the command directly as that will prevent
-# the container from exiting when `docker stop` is run.
-# See https://stackoverflow.com/a/55678435
-exec {log_fd}> >(exec tee $REPO_DIR/.jupyter-server-log.txt)
+# goals:
+# - load environment variables from a login shell (bash -l)
+# - preserve signal handling of subprocess (kill -TERM and friends)
+# - tee output to a log file
 
-if [[ ! -z "${R2D_ENTRYPOINT:-}" ]]; then
-    if [[ ! -x "$R2D_ENTRYPOINT" ]]; then
-        chmod u+x "$R2D_ENTRYPOINT"
-    fi
-    exec "$R2D_ENTRYPOINT" "$@" 2>&1 >&"$log_fd"
-else
-    exec "$@" 2>&1 >&"$log_fd"
-fi
+import fcntl
+import os
+import select
+import signal
+import subprocess
+import sys
 
-# Close the logging output again
-exec {log_fd}>&-
+# output chunk size to read
+CHUNK_SIZE = 1024
+
+# signals to be forwarded to the child
+# everything catchable, excluding SIGCHLD
+SIGNALS = set(signal.Signals) - {signal.SIGKILL, signal.SIGSTOP, signal.SIGCHLD}
+
+
+def main():
+
+    # open log file to send output
+    log_file = open(
+        os.path.join(os.environ.get("REPO_DIR", "."), ".jupyter-server-log.txt"),
+        "ab",
+    )
+
+    # build the command
+    # like `exec "$@"`
+    command = sys.argv[1:]
+    # load entrypoint override from env
+    r2d_entrypoint = os.environ.get("R2D_ENTRYPOINT")
+    if r2d_entrypoint:
+        command.insert(0, r2d_entrypoint)
+
+    # launch the subprocess
+    child = subprocess.Popen(
+        command,
+        bufsize=1,
+        stdout=subprocess.PIPE,
+        stderr=subprocess.STDOUT,
+    )
+
+    # hook up ~all signals so that every signal the parent gets,
+    # the children also get
+
+    def relay_signal(sig, frame):
+        """Relay a signal to children"""
+        # DEBUG: show signal
+        child.send_signal(sig)
+
+    for signum in SIGNALS:
+        signal.signal(signum, relay_signal)
+
+    # tee output from child to both our stdout and the log file
+    def tee(chunk):
+        """Tee output from child to both our stdout and the log file"""
+        for f in [sys.stdout.buffer, log_file]:
+            f.write(chunk)
+            f.flush()
+
+    # make stdout pipe non-blocking
+    # this means child.stdout.read(nbytes)
+    # will always return immediately, even if there's nothing to read
+    flags = fcntl.fcntl(child.stdout, fcntl.F_GETFL)
+    fcntl.fcntl(child.stdout, fcntl.F_SETFL, flags | os.O_NONBLOCK)
+    poller = select.poll()
+    poller.register(child.stdout)
+
+    # while child is running, constantly relay output
+    while child.poll() is None:
+        chunk = child.stdout.read(CHUNK_SIZE)
+        if chunk:
+            tee(chunk)
+        else:
+            # empty chunk means nothing to read
+            # wait for output on the pipe
+            # timeout is in milliseconds
+            poller.poll(1000)
+
+    # child has exited, continue relaying any remaining output
+    # At this point, read() will return an empty string when it's done
+    chunk = child.stdout.read()
+    while chunk:
+        tee(chunk)
+        chunk = child.stdout.read()
+
+    # make our returncode match the child's returncode
+    sys.exit(child.returncode)
+
+
+if __name__ == "__main__":
+    main()

tests/unit/test_env.py

@@ -3,12 +3,13 @@ Test that environment variables may be defined
 """
 import os
 import subprocess
+import sys
 import tempfile
 import time
 from getpass import getuser
 
 
-def test_env():
+def test_env(capfd):
     """
     Validate that you can define environment variables
 
@@ -42,20 +43,19 @@ def test_env():
                 # value
                 "--env",
                 "SPAM_2=",
-                # "--",
                 tmpdir,
+                "--",
                 "/bin/bash",
                 "-c",
                 # Docker exports all passed env variables, so we can
                 # just look at exported variables.
-                "export; sleep 1",
-                # "export; echo TIMDONE",
-                # "export",
+                "export",
             ],
-            universal_newlines=True,
-            stdout=subprocess.PIPE,
-            stderr=subprocess.PIPE,
         )
+    captured = capfd.readouterr()
+    print(captured.out, end="")
+    print(captured.err, file=sys.stderr, end="")
+
     assert result.returncode == 0
 
     # all docker output is returned by repo2docker on stderr
@@ -63,11 +63,8 @@ def test_env():
     # stdout should be empty
     assert not result.stdout
 
-    print(result.stderr.split("\n"))
-    # assert False
-
     # stderr should contain lines of output
-    declares = [x for x in result.stderr.split("\n") if x.startswith("declare")]
+    declares = [x for x in captured.err.splitlines() if x.startswith("declare")]
     assert 'declare -x FOO="{}"'.format(ts) in declares
     assert 'declare -x BAR="baz"' in declares
     assert 'declare -x SPAM="eggs"' in declares