[jboss-jira] [JBoss JIRA] (JGRP-2249) CENTRAL_LOCK2: reconciliation protocol on coord change

[Bela Ban (JIRA)]

     [ https://issues.jboss.org/browse/JGRP-2249?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Bela Ban resolved JGRP-2249.
----------------------------
    Resolution: Done


Done, design doc is [1].

Next step is to think whether distributed conditions really make sense, unless we have distributed state, too. See my email below:

I'm implementing a new version of the locking protocol, CENTRAL_LOCK2, and am currently looking at conditions.

Since you're the author of the Condition part of the LockService...

Do you think distributed conditions make sense?

Usually, a condition is associated with some state, and the await()/signal() methods check/modify that state.

However, to be truly distributed, that state would have to be present in every member, and be changed in every member accordingly, too.

Do you assume these state changes (e.g. before a signal()) happen outside of the scope of the LockService, e.g. via messages/RPCs?

I'm think whether we should pull support for conditions, ie. Lock.newCondition() would throw an UnsupportedOperationException.

How do you guys handle this in the Infinispan locking service (Pedro)?

WDYT?


[1] https://github.com/belaban/JGroups/blob/master/doc/design/CENTRAL_LOCK2.txt

> CENTRAL_LOCK2: reconciliation protocol on coord change
> ------------------------------------------------------
>
>                 Key: JGRP-2249
>                 URL: https://issues.jboss.org/browse/JGRP-2249
>             Project: JGroups
>          Issue Type: Feature Request
>            Reporter: Bela Ban
>            Assignee: Bela Ban
>             Fix For: 4.0.13
>
>
> Implement a reconciliation protocol when the coord changes: instead of backing up all lock information to another member, the new coord asks all members for their current lock information (acquired locks, pending acquire and release requests) and constructs the lock table accordingly.
> This is described in https://issues.jboss.org/browse/JGRP-2234 (later comments).
> Comment from 2234:
> Clients need to have the following information:
> * Locks they acquired
> * Pending lock requests; locks which they want to acquire but for which they haven't yet received a LOCK_GRANTED response
> * Pending lock release requests; lock that have been released, but for which no RELEASE_LOCK_OK response has been received
> * Ditto for conditions, but we'll tackle them in a second stage
> The reconciliation protocol queues all new requests on the coord and asks all members for their lock information. Once the coord has received this information from all members, it applies this and then drains the queue of pending requests.
> It is important that the requests are ordered per member, ie. a release(L) cannot come before a lock(L).
> Since {{CENTRAL_LOCK}} allows for multiple members to hold the same lock in a split brain scenario, we need to think about how to handle merging where the coord detects that multiple members hold the same lock...



--
This message was sent by Atlassian JIRA
(v7.5.0#75005)