Also, reverts r2197, as it is unclear what bug it was trying to fix, and it
doens't seem required (there is no call to Replicator.act() outside of
doOperation).

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2223 71dcc9de-d417-0410-9af5-da40c76e7ee4

This causes garbage collection problems, as BaseConnection instances also
hold a reference to HandlerSwitcher instance. It is unclear why it fails,
as microbenchs show this should not be a problem for gc.
Also, move loggers out of HandlerSwitcher.setHandler to avoid passing a
connection parameter just for them.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2222 71dcc9de-d417-0410-9af5-da40c76e7ee4

It is more efficient to provide a boundary value than a row count range.
This fixes replication on partitions with a large number of objects, revisions
or transactions: query time is now constant where it used to increase, causing
timeout problems when query duration exceeded ping time + ping timeout (11s
currently).

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2221 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2220 71dcc9de-d417-0410-9af5-da40c76e7ee4

Previous name was too close to the regular-user SQL connection getter.
Also, prevent subclasses from calling it.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2219 71dcc9de-d417-0410-9af5-da40c76e7ee4

This made it impossible to detect when a storage didn't contain an object.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2218 71dcc9de-d417-0410-9af5-da40c76e7ee4

It is not possible to move them out of container loop, as tid is a loop
variable.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2217 71dcc9de-d417-0410-9af5-da40c76e7ee4

Also, stop using a dict in an inefficient way, and replace it by a set.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2216 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2215 71dcc9de-d417-0410-9af5-da40c76e7ee4

Also, factorises packet creation.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2214 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2213 71dcc9de-d417-0410-9af5-da40c76e7ee4

onConnectionClosed is already called inside Connection.close() when peer
is identified (via setOnClose). Calling it here would trigger an exception
as it verifies that _connection is set before un-setting it.
As a replacement, verify that _connection is not set when
Connection.close() returns (so onConnectionClosed was actually called).

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2212 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2211 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2210 71dcc9de-d417-0410-9af5-da40c76e7ee4

It is possible for rid not to be in partition_dict or new_partition_dict,
as this method is called from a master notification (partition changed),
which might happen outside of any replication (so when both dicts are
empty).

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2209 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2208 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2207 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2206 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2205 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2204 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2203 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2202 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2201 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2200 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2199 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2198 71dcc9de-d417-0410-9af5-da40c76e7ee4

The replicator must be stopped to avoid storages request unfinished
transactions to the master while it is not able to answer. This prevent
from considering the storage as broken.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2197 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2196 71dcc9de-d417-0410-9af5-da40c76e7ee4

If a client connect to a storage more than one time, cut the old connection
and replace it by the new one. This is required if the client reconnect to
fast and the storage handle in single poll loop the new connection *then*
the lost of the previous connection.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2195 71dcc9de-d417-0410-9af5-da40c76e7ee4

The only purpose on this is to have the log in setHandler() that display the
connection's UUID instead of None.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2194 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2193 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2192 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2191 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2190 71dcc9de-d417-0410-9af5-da40c76e7ee4

Avoid duplicates in deployment configurations.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2189 71dcc9de-d417-0410-9af5-da40c76e7ee4

With a sample configuration for supervisor.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2188 71dcc9de-d417-0410-9af5-da40c76e7ee4

About the potential use of twisted framework.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2187 71dcc9de-d417-0410-9af5-da40c76e7ee4

Since the last OID used (in a transaction) is synchronized with the storages,
it is no more required to restart a NEO cluster after an import. Remove the
related message.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2186 71dcc9de-d417-0410-9af5-da40c76e7ee4

- A storage is not operational while connecting to the primary master
- A storage is not ready until connected to a primary master, with an
operational partition table and fully initialized. The 'ready' attribute
must be initialized to False when PrimaryFailure or OperationFailure
exceptions are reaised to prevent acceptation of incoming clients (see
storage/handlers/identification.py that check app.ready value)

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2185 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@2184 71dcc9de-d417-0410-9af5-da40c76e7ee4