git-svn-id: https://svn.erp5.org/repos/neo/trunk@2612 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2611 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2610 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2609 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2608 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2607 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2606 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2605 71dcc9de-d417-0410-9af5-da40c76e7ee4

With conflict resolution, multiple stores can happen and register more than
once a event with a given key. Avoid crash on a KeyError exception when
removing the key from queued events.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2604 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2603 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2602 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2601 71dcc9de-d417-0410-9af5-da40c76e7ee4

Detect if a client requests multiple undo/store based on wrong revisions.
Also, don't try to acquire lock when we know we have it.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2600 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2599 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2598 71dcc9de-d417-0410-9af5-da40c76e7ee4

This is important for chaining "undo" for a given object in a single
transaction, to allow checking for conflicts.
Also, this means that get_baseTID return value cannot be used to update
cache when undoing, so only update cache when storing a new revision.
Actually, cache was never updated when undoing anyway, as current snapshot
TID was unlikely to exist as a serial for any object, so it would not be
found in cache and update would be a no-op.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2597 71dcc9de-d417-0410-9af5-da40c76e7ee4

Also reverts r2533: even if it's true that TIDs have no meaning at that level,
they are a handy way to prioritise a transaction over another, to break lock
cycles (aka deadlocks). This is the "detection" part of the change.
When a storage reports a deadlock, client must store all already-stored
objects again with "unlock" flag set.
Upon receiving those store requests, storage must release locks held by
transaction on those objects, and requeue the store request. If client didn't
hold any lock (initial store was still in queue), drop the second store
request.
This doesn't solve possible deadlocks if ZODB-level sends us objects in a
different order (ex: client 1 sending [OID1, OID2] & client 2 sending
[OID2, OID1]).
There is one important change to r2533's revert, which queues older
transactions and notifies deadlock for younger ones. The original code did it
the other way around, and it looks unfair to old transactions (they will keep
being delayed by younger ones, and will just get older and older).

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2596 71dcc9de-d417-0410-9af5-da40c76e7ee4

If a conflict gets reported while handling a previously-reported conflict
on the same object, second conflict is dropped on
  conflict_serial_dict.pop()
before it is handled.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2595 71dcc9de-d417-0410-9af5-da40c76e7ee4

This will allow sending requests even when no conflict occur. For example,
it can happen to avoid a deadlock on storage nodes (store again, no conflict
to resolve).
Also, call conflict resolution as soon as a conflict is known, to reduce
tpc_vote duration.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2594 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2593 71dcc9de-d417-0410-9af5-da40c76e7ee4

This was removed accidentally in r2114.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2592 71dcc9de-d417-0410-9af5-da40c76e7ee4

Don't block on a non-ready node but do not skip it
Fix test method name and update it (it will never raise StopIteration)

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2591 71dcc9de-d417-0410-9af5-da40c76e7ee4

A serial is just like a tid.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2590 71dcc9de-d417-0410-9af5-da40c76e7ee4

And remove its definition from protocol as it's used only by tests.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2589 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2588 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2587 71dcc9de-d417-0410-9af5-da40c76e7ee4

- Runner do not define the environment, this is under the responsibility
of the parent script.
- Prefix all environment variables with NEO_TEST to avoid conflicts with
other applications
- Runner display all NEO-related environment variables in the report
- Cosmetic change: increase benchmark runner status column size

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2586 71dcc9de-d417-0410-9af5-da40c76e7ee4

Add a test to ensure the admin is notified of connected/disconnected
nodes during recovery.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2585 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2584 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2583 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2582 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2581 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2580 71dcc9de-d417-0410-9af5-da40c76e7ee4

Remove 'else' sections of 'for' loops from app.
(and the one in checkCurrentSerialInTransaction was broken)

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2579 71dcc9de-d417-0410-9af5-da40c76e7ee4

This commit fix random issues found with functionnal tests where the client
was refuse by the storage, because the latter was not fully initialized,
but never tried to reconnect to it if no other storages were available.

The main change introoduced is the availability of 'iterateForObject'
method on ConnectionPool. It allow iterate over potential node connections
for a given object id with the ability of waiting for the node to be ready
if not. It includes the common pattern that retreive the cell list,
randomize then sort them and never returns a None value, which suppose that
the outer loop must check if at least one iteration happens, for example.

Also included:
- getPartitionTable is now private because the connection needs it
- Deletion of _getCellListFor*
- Fixed tests
- New tests for ConnectionPool.iterateForObject

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2578 71dcc9de-d417-0410-9af5-da40c76e7ee4

So that search and add in connection_dict is made in the same method.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2577 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2576 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2575 71dcc9de-d417-0410-9af5-da40c76e7ee4

This fixes cases where storage tries to answer a request after executing
queued events. This cannto happen with current code, but will happen in
near future.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2574 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2573 71dcc9de-d417-0410-9af5-da40c76e7ee4