问题
I have a python script which is querying a MySQL server on a shared linux host. For some reason, queries to MySQL often return a "server has gone away" error:
_mysql_exceptions.OperationalError: (2006, 'MySQL server has gone away')
If you try the query again immediately afterwards, it usually succeeds. So, I'd like to know if there's a sensible way in python to try to execute a query, and if it fails, to try again, up to a fixed number of tries. Probably I'd want it to try 5 times before giving up altogether.
Here's the kind of code I have:
conn = MySQLdb.connect(host, user, password, database)
cursor = conn.cursor()
try:
cursor.execute(query)
rows = cursor.fetchall()
for row in rows:
# do something with the data
except MySQLdb.Error, e:
print "MySQL Error %d: %s" % (e.args[0], e.args[1])
Clearly I could do it by having another attempt in the except clause, but that's incredibly ugly, and I have a feeling there must be a decent way to achieve this.
回答1:
How about:
conn = MySQLdb.connect(host, user, password, database)
cursor = conn.cursor()
attempts = 0
while attempts < 3:
try:
cursor.execute(query)
rows = cursor.fetchall()
for row in rows:
# do something with the data
break
except MySQLdb.Error, e:
attempts += 1
print "MySQL Error %d: %s" % (e.args[0], e.args[1])
回答2:
Building on Dana's answer, you might want to do this as a decorator:
def retry(howmany):
def tryIt(func):
def f():
attempts = 0
while attempts < howmany:
try:
return func()
except:
attempts += 1
return f
return tryIt
Then...
@retry(5)
def the_db_func():
# [...]
Enhanced version that uses the decorator module
import decorator, time
def retry(howmany, *exception_types, **kwargs):
timeout = kwargs.get('timeout', 0.0) # seconds
@decorator.decorator
def tryIt(func, *fargs, **fkwargs):
for _ in xrange(howmany):
try: return func(*fargs, **fkwargs)
except exception_types or Exception:
if timeout is not None: time.sleep(timeout)
return tryIt
Then...
@retry(5, MySQLdb.Error, timeout=0.5)
def the_db_func():
# [...]
To install the decorator module:
$ easy_install decorator
回答3:
UPDATE: there is a better maintained fork of the retrying library called tenacity, which supports more features and is in general more flexible.
Yes, there is the retrying library, which has a decorator that implements several kinds of retrying logic that you can combine:
Some examples:
@retry(stop_max_attempt_number=7)
def stop_after_7_attempts():
print "Stopping after 7 attempts"
@retry(wait_fixed=2000)
def wait_2_s():
print "Wait 2 second between retries"
@retry(wait_exponential_multiplier=1000, wait_exponential_max=10000)
def wait_exponential_1000():
print "Wait 2^x * 1000 milliseconds between each retry,"
print "up to 10 seconds, then 10 seconds afterwards"
回答4:
conn = MySQLdb.connect(host, user, password, database)
cursor = conn.cursor()
for i in range(3):
try:
cursor.execute(query)
rows = cursor.fetchall()
for row in rows:
# do something with the data
break
except MySQLdb.Error, e:
print "MySQL Error %d: %s" % (e.args[0], e.args[1])
回答5:
I'd refactor it like so:
def callee(cursor):
cursor.execute(query)
rows = cursor.fetchall()
for row in rows:
# do something with the data
def caller(attempt_count=3, wait_interval=20):
""":param wait_interval: In seconds."""
conn = MySQLdb.connect(host, user, password, database)
cursor = conn.cursor()
for attempt_number in range(attempt_count):
try:
callee(cursor)
except MySQLdb.Error, e:
logging.warn("MySQL Error %d: %s", e.args[0], e.args[1])
time.sleep(wait_interval)
else:
break
Factoring out the callee function seems to break up the functionality so that it's easy to see the business logic without getting bogged down in the retry code.
回答6:
Like S.Lott, I like a flag to check if we're done:
conn = MySQLdb.connect(host, user, password, database)
cursor = conn.cursor()
success = False
attempts = 0
while attempts < 3 and not success:
try:
cursor.execute(query)
rows = cursor.fetchall()
for row in rows:
# do something with the data
success = True
except MySQLdb.Error, e:
print "MySQL Error %d: %s" % (e.args[0], e.args[1])
attempts += 1
回答7:
def successful_transaction(transaction):
try:
transaction()
return True
except SQL...:
return False
succeeded = any(successful_transaction(transaction)
for transaction in repeat(transaction, 3))
回答8:
1.Definition:
def try_three_times(express):
att = 0
while att < 3:
try: return express()
except: att += 1
else: return u"FAILED"
2.Usage:
try_three_times(lambda: do_some_function_or_express())
I use it for parse html context.
回答9:
This is my generic solution:
class TryTimes(object):
''' A context-managed coroutine that returns True until a number of tries have been reached. '''
def __init__(self, times):
''' times: Number of retries before failing. '''
self.times = times
self.count = 0
def __next__(self):
''' A generator expression that counts up to times. '''
while self.count < self.times:
self.count += 1
yield False
def __call__(self, *args, **kwargs):
''' This allows "o() calls for "o = TryTimes(3)". '''
return self.__next__().next()
def __enter__(self):
''' Context manager entry, bound to t in "with TryTimes(3) as t" '''
return self
def __exit__(self, exc_type, exc_val, exc_tb):
''' Context manager exit. '''
return False # don't suppress exception
This allows code like the following:
with TryTimes(3) as t:
while t():
print "Your code to try several times"
Also possible:
t = TryTimes(3)
while t():
print "Your code to try several times"
This can be improved by handling exceptions in a more intuitive way, I hope. Open to suggestions.
来源：https://stackoverflow.com/questions/567622/is-there-a-pythonic-way-to-try-something-up-to-a-maximum-number-of-times