Comment 0 for bug 1975439

The script is:
```python
from tooz import coordination
from tooz import _retry
from tooz.drivers import memcached as mem1
from oslo_utils import timeutils
import uuid
import time
import tenacity
from tenacity import *
import logging
import threading
import functools
import pdb

LOG = logging.getLogger(__name__)

pdb.set_trace()

LOCK_TIMEOUT = 30
url = "memcached://100.101.89.222:11211"
member_id = ("cinder-" + str(uuid.uuid4())).encode('ascii')
coordinator = coordination.get_coordinator(
    url, member_id, lock_timeout=LOCK_TIMEOUT
)
coordinator.start(start_heart=True)

lock_name = "uuid-delete_snasphot".encode("ascii")

# to get a lock
lock = coordinator.get_lock(lock_name)

@_retry.retry(stop_max_delay=True)
@mem1._translate_failures
def _acquire():
    print("trying add lock %s", lock.name)
    if lock.coord.client.add(
            lock.name,
            lock.coord._member_id,
            expire=LOCK_TIMEOUT,
            noreply=False):
        print("add lock success")
        return True
    raise _retry.TryAgain

# to acquire lock
_acquire()

lock.coord.client.delete(lock.name)
lock.release()
coordinator.stop()

```

when lock was got and was about to acquire lock, i shutdown the memcache server '100.101.89.222:11211', It never tried acquire again.

the result is:
```bash
(Pdb)
> /home/lz/tes-memcache.py(24)<module>()
-> coordinator = coordination.get_coordinator(
(Pdb)
> /home/lz/tes-memcache.py(25)<module>()
-> url, member_id, lock_timeout=LOCK_TIMEOUT
(Pdb)
> /home/lz/tes-memcache.py(27)<module>()
-> coordinator.start(start_heart=True)
(Pdb)
> /home/lz/tes-memcache.py(29)<module>()
-> lock_name = "uuid-delete_snasphot".encode("ascii")
(Pdb) c
trying add lock %s b'__TOOZ_LOCK_uuid-delete_snasphot'
Traceback (most recent call last):
  File "/home/lz/my_project_env/lib/python3.6/site-packages/tooz/drivers/memcached.py", line 43, in _failure_translator
    yield
  File "/home/lz/my_project_env/lib/python3.6/site-packages/tooz/drivers/memcached.py", line 70, in wrapper
    return func(*args, **kwargs)
  File "tes-memcache.py", line 47, in _acquire
    noreply=False):
  File "/home/lz/my_project_env/lib/python3.6/site-packages/pymemcache/client/base.py", line 1169, in add
    flags=flags)
  File "/home/lz/my_project_env/lib/python3.6/site-packages/pymemcache/client/base.py", line 393, in add
    flags=flags)[key]
  File "/home/lz/my_project_env/lib/python3.6/site-packages/pymemcache/client/base.py", line 942, in _store_cmd
    buf, line = _readline(self.sock, buf)
  File "/home/lz/my_project_env/lib/python3.6/site-packages/pymemcache/client/base.py", line 1266, in _readline
    raise MemcacheUnexpectedCloseError()
pymemcache.exceptions.MemcacheUnexpectedCloseError
```

Is retry strategy "retry_if_exception_type" more proper here other than "retry_never"?