procutil: always waiting on child processes to prevent zombies with 'hg serve'
When runbgcommand is invoked by an extension with ensurestart=False, we never
called waitpid - which is fine in most cases, except if that's happening on a
command server (e.g. chg), in which case the child defunct process will just
sit there for as long as the server is running.
The actual semantics of SIGCHLD signal handling is a lot more complex than
it seems, and the POSIX standard *seems* to read that it's ignored by default
and everything would just work without the waitpid if we're not listening for
it, but the truth is that it's only ignored if we *explicitly* set it to
SIG_IGN. We further cannot set it to SIG_IGN or to a catch-all handler across
all of 'hg serve', because Python's suprocess.Popen relies on that signal,
and a few specific parts of hg also set custom handlers, so instead we wait
for specific PIDs in dedicated threads.
I did a poor-man's benchmark of the thread creation and it seems to take
about 1ms, which is way better than the 20+ms from ensurestart=True.
Differential Revision: https://phab.mercurial-scm.org/D8497
# Infinite push
#
# Copyright 2016 Facebook, Inc.
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import
import logging
import os
import time
import warnings
import mysql.connector
from mercurial import pycompat
from . import indexapi
def _convertbookmarkpattern(pattern):
pattern = pattern.replace(b'_', b'\\_')
pattern = pattern.replace(b'%', b'\\%')
if pattern.endswith(b'*'):
pattern = pattern[:-1] + b'%'
return pattern
class sqlindexapi(indexapi.indexapi):
'''
Sql backend for infinitepush index. See schema.sql
'''
def __init__(
self,
reponame,
host,
port,
database,
user,
password,
logfile,
loglevel,
waittimeout=300,
locktimeout=120,
):
super(sqlindexapi, self).__init__()
self.reponame = reponame
self.sqlargs = {
b'host': host,
b'port': port,
b'database': database,
b'user': user,
b'password': password,
}
self.sqlconn = None
self.sqlcursor = None
if not logfile:
logfile = os.devnull
logging.basicConfig(filename=logfile)
self.log = logging.getLogger()
self.log.setLevel(loglevel)
self._connected = False
self._waittimeout = waittimeout
self._locktimeout = locktimeout
def sqlconnect(self):
if self.sqlconn:
raise indexapi.indexexception(b"SQL connection already open")
if self.sqlcursor:
raise indexapi.indexexception(
b"SQL cursor already open without connection"
)
retry = 3
while True:
try:
self.sqlconn = mysql.connector.connect(**self.sqlargs)
# Code is copy-pasted from hgsql. Bug fixes need to be
# back-ported!
# The default behavior is to return byte arrays, when we
# need strings. This custom convert returns strings.
self.sqlconn.set_converter_class(CustomConverter)
self.sqlconn.autocommit = False
break
except mysql.connector.errors.Error:
# mysql can be flakey occasionally, so do some minimal
# retrying.
retry -= 1
if retry == 0:
raise
time.sleep(0.2)
waittimeout = self.sqlconn.converter.escape(b'%s' % self._waittimeout)
self.sqlcursor = self.sqlconn.cursor()
self.sqlcursor.execute(b"SET wait_timeout=%s" % waittimeout)
self.sqlcursor.execute(
b"SET innodb_lock_wait_timeout=%s" % self._locktimeout
)
self._connected = True
def close(self):
"""Cleans up the metadata store connection."""
with warnings.catch_warnings():
warnings.simplefilter(b"ignore")
self.sqlcursor.close()
self.sqlconn.close()
self.sqlcursor = None
self.sqlconn = None
def __enter__(self):
if not self._connected:
self.sqlconnect()
return self
def __exit__(self, exc_type, exc_val, exc_tb):
if exc_type is None:
self.sqlconn.commit()
else:
self.sqlconn.rollback()
def addbundle(self, bundleid, nodesctx):
if not self._connected:
self.sqlconnect()
self.log.info(b"ADD BUNDLE %r %r" % (self.reponame, bundleid))
self.sqlcursor.execute(
b"INSERT INTO bundles(bundle, reponame) VALUES (%s, %s)",
params=(bundleid, self.reponame),
)
for ctx in nodesctx:
self.sqlcursor.execute(
b"INSERT INTO nodestobundle(node, bundle, reponame) "
b"VALUES (%s, %s, %s) ON DUPLICATE KEY UPDATE "
b"bundle=VALUES(bundle)",
params=(ctx.hex(), bundleid, self.reponame),
)
extra = ctx.extra()
author_name = ctx.user()
committer_name = extra.get(b'committer', ctx.user())
author_date = int(ctx.date()[0])
committer_date = int(extra.get(b'committer_date', author_date))
self.sqlcursor.execute(
b"INSERT IGNORE INTO nodesmetadata(node, message, p1, p2, "
b"author, committer, author_date, committer_date, "
b"reponame) VALUES "
b"(%s, %s, %s, %s, %s, %s, %s, %s, %s)",
params=(
ctx.hex(),
ctx.description(),
ctx.p1().hex(),
ctx.p2().hex(),
author_name,
committer_name,
author_date,
committer_date,
self.reponame,
),
)
def addbookmark(self, bookmark, node):
"""Takes a bookmark name and hash, and records mapping in the metadata
store."""
if not self._connected:
self.sqlconnect()
self.log.info(
b"ADD BOOKMARKS %r bookmark: %r node: %r"
% (self.reponame, bookmark, node)
)
self.sqlcursor.execute(
b"INSERT INTO bookmarkstonode(bookmark, node, reponame) "
b"VALUES (%s, %s, %s) ON DUPLICATE KEY UPDATE node=VALUES(node)",
params=(bookmark, node, self.reponame),
)
def addmanybookmarks(self, bookmarks):
if not self._connected:
self.sqlconnect()
args = []
values = []
for bookmark, node in pycompat.iteritems(bookmarks):
args.append(b'(%s, %s, %s)')
values.extend((bookmark, node, self.reponame))
args = b','.join(args)
self.sqlcursor.execute(
b"INSERT INTO bookmarkstonode(bookmark, node, reponame) "
b"VALUES %s ON DUPLICATE KEY UPDATE node=VALUES(node)" % args,
params=values,
)
def deletebookmarks(self, patterns):
"""Accepts list of bookmark patterns and deletes them.
If `commit` is set then bookmark will actually be deleted. Otherwise
deletion will be delayed until the end of transaction.
"""
if not self._connected:
self.sqlconnect()
self.log.info(b"DELETE BOOKMARKS: %s" % patterns)
for pattern in patterns:
pattern = _convertbookmarkpattern(pattern)
self.sqlcursor.execute(
b"DELETE from bookmarkstonode WHERE bookmark LIKE (%s) "
b"and reponame = %s",
params=(pattern, self.reponame),
)
def getbundle(self, node):
"""Returns the bundleid for the bundle that contains the given node."""
if not self._connected:
self.sqlconnect()
self.log.info(b"GET BUNDLE %r %r" % (self.reponame, node))
self.sqlcursor.execute(
b"SELECT bundle from nodestobundle "
b"WHERE node = %s AND reponame = %s",
params=(node, self.reponame),
)
result = self.sqlcursor.fetchall()
if len(result) != 1 or len(result[0]) != 1:
self.log.info(b"No matching node")
return None
bundle = result[0][0]
self.log.info(b"Found bundle %r" % bundle)
return bundle
def getnode(self, bookmark):
"""Returns the node for the given bookmark. None if it doesn't exist."""
if not self._connected:
self.sqlconnect()
self.log.info(
b"GET NODE reponame: %r bookmark: %r" % (self.reponame, bookmark)
)
self.sqlcursor.execute(
b"SELECT node from bookmarkstonode WHERE "
b"bookmark = %s AND reponame = %s",
params=(bookmark, self.reponame),
)
result = self.sqlcursor.fetchall()
if len(result) != 1 or len(result[0]) != 1:
self.log.info(b"No matching bookmark")
return None
node = result[0][0]
self.log.info(b"Found node %r" % node)
return node
def getbookmarks(self, query):
if not self._connected:
self.sqlconnect()
self.log.info(
b"QUERY BOOKMARKS reponame: %r query: %r" % (self.reponame, query)
)
query = _convertbookmarkpattern(query)
self.sqlcursor.execute(
b"SELECT bookmark, node from bookmarkstonode WHERE "
b"reponame = %s AND bookmark LIKE %s",
params=(self.reponame, query),
)
result = self.sqlcursor.fetchall()
bookmarks = {}
for row in result:
if len(row) != 2:
self.log.info(b"Bad row returned: %s" % row)
continue
bookmarks[row[0]] = row[1]
return bookmarks
def saveoptionaljsonmetadata(self, node, jsonmetadata):
if not self._connected:
self.sqlconnect()
self.log.info(
(
b"INSERT METADATA, QUERY BOOKMARKS reponame: %r "
+ b"node: %r, jsonmetadata: %s"
)
% (self.reponame, node, jsonmetadata)
)
self.sqlcursor.execute(
b"UPDATE nodesmetadata SET optional_json_metadata=%s WHERE "
b"reponame=%s AND node=%s",
params=(jsonmetadata, self.reponame, node),
)
class CustomConverter(mysql.connector.conversion.MySQLConverter):
"""Ensure that all values being returned are returned as python string
(versus the default byte arrays)."""
def _STRING_to_python(self, value, dsc=None):
return str(value)
def _VAR_STRING_to_python(self, value, dsc=None):
return str(value)
def _BLOB_to_python(self, value, dsc=None):
return str(value)