Commit a5b360e4 authored by Michael Hanselmann's avatar Michael Hanselmann

Re-wrap locking changes design to 76 chars per line

Signed-off-by: default avatarMichael Hanselmann <hansmi@google.com>
Reviewed-by: default avatarIustin Pop <iustin@google.com>
parent 700bb843
......@@ -87,24 +87,26 @@ Locking improvements
Current State and shortcomings
++++++++++++++++++++++++++++++
The class ``LockSet`` (see ``lib/locking.py``) is a container for one or many
``SharedLock`` instances. It provides an interface to add/remove locks and to
acquire and subsequently release any number of those locks contained in it.
The class ``LockSet`` (see ``lib/locking.py``) is a container for one or
many ``SharedLock`` instances. It provides an interface to add/remove locks
and to acquire and subsequently release any number of those locks contained
in it.
Locks in a ``LockSet`` are always acquired in alphabetic order. Due to the way
we're using locks for nodes and instances (the single cluster lock isn't
Locks in a ``LockSet`` are always acquired in alphabetic order. Due to the
way we're using locks for nodes and instances (the single cluster lock isn't
affected by this issue) this can lead to long delays when acquiring locks if
another operation tries to acquire multiple locks but has to wait for yet
another operation.
In the following demonstration we assume to have the instance locks ``inst1``,
``inst2``, ``inst3`` and ``inst4``.
In the following demonstration we assume to have the instance locks
``inst1``, ``inst2``, ``inst3`` and ``inst4``.
#. Operation A grabs lock for instance ``inst4``.
#. Operation B wants to acquire all instance locks in alphabetic order, but it
has to wait for ``inst4``.
#. Operation B wants to acquire all instance locks in alphabetic order, but
it has to wait for ``inst4``.
#. Operation C tries to lock ``inst1``, but it has to wait until
Operation B (which is trying to acquire all locks) releases the lock again.
Operation B (which is trying to acquire all locks) releases the lock
again.
#. Operation A finishes and releases lock on ``inst4``. Operation B can
continue and eventually releases all locks.
#. Operation C can get ``inst1`` lock and finishes.
......@@ -123,22 +125,22 @@ Acquiring locks for OpCode execution is always done in blocking mode. They
won't return until the lock has successfully been acquired (or an error
occurred, although we won't cover that case here).
``SharedLock`` and ``LockSet`` must be able to be acquired in a
non-blocking way. They must support a timeout and abort trying to acquire
the lock(s) after the specified amount of time.
``SharedLock`` and ``LockSet`` must be able to be acquired in a non-blocking
way. They must support a timeout and abort trying to acquire the lock(s)
after the specified amount of time.
Retry acquiring locks
^^^^^^^^^^^^^^^^^^^^^
To prevent other operations from waiting for a long time, such as described in
the demonstration before, ``LockSet`` must not keep locks for a prolonged period
of time when trying to acquire two or more locks. Instead it should, with an
increasing timeout for acquiring all locks, release all locks again and
sleep some time if it fails to acquire all requested locks.
To prevent other operations from waiting for a long time, such as described
in the demonstration before, ``LockSet`` must not keep locks for a prolonged
period of time when trying to acquire two or more locks. Instead it should,
with an increasing timeout for acquiring all locks, release all locks again
and sleep some time if it fails to acquire all requested locks.
A good timeout value needs to be determined. In any case should ``LockSet``
proceed to acquire locks in blocking mode after a few (unsuccessful) attempts
to acquire all requested locks.
proceed to acquire locks in blocking mode after a few (unsuccessful)
attempts to acquire all requested locks.
One proposal for the timeout is to use ``2**tries`` seconds, where ``tries``
is the number of unsuccessful tries.
......@@ -151,13 +153,13 @@ Other solutions discussed
+++++++++++++++++++++++++
There was also some discussion on going one step further and extend the job
queue (see ``lib/jqueue.py``) to select the next task for a worker depending on
whether it can acquire the necessary locks. While this may reduce the number of
necessary worker threads and/or increase throughput on large clusters with many
jobs, it also brings many potential problems, such as contention and increased
memory usage, with it. As this would be an extension of the changes proposed
before it could be implemented at a later point in time, but we decided to stay
with the simpler solution for now.
queue (see ``lib/jqueue.py``) to select the next task for a worker depending
on whether it can acquire the necessary locks. While this may reduce the
number of necessary worker threads and/or increase throughput on large
clusters with many jobs, it also brings many potential problems, such as
contention and increased memory usage, with it. As this would be an
extension of the changes proposed before it could be implemented at a later
point in time, but we decided to stay with the simpler solution for now.
Feature changes
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment