[
https://issues.jboss.org/browse/WFLY-1416?page=com.atlassian.jira.plugin....
]
Richard Achmatowicz commented on WFLY-1416:
-------------------------------------------
The results of running probe.sh on the first example above:
{noformat}
[nrla@lenovo bin]$ ./probe.sh -addr 224.0.75.75 op=RELAY2.printRoutes
-- send probe on /224.0.75.75:7500
#1 (211 bytes):
local_addr=LON-1/web [54877f5a-7378-f542-fec5-8b89af0b6d89]
cluster=web
RELAY2.printRoutes=n/a (not site master)
physical_addr=127.0.0.1:55300
view=[LON-0/web|1] [LON-0/web, LON-1/web]
version=3.3.0.Final
#2 (274 bytes):
local_addr=LON-0/web [a033d27d-6af1-3ec7-7ee2-3241e4e62978]
cluster=web
RELAY2.printRoutes=LON --> _LON-0/web:LON [UP]
NYC --> _NYC-0/web:NYC [UP]
SFO --> _SFO-0/web:SFO [UP]
physical_addr=127.0.0.1:55200
view=[LON-0/web|1] [LON-0/web, LON-1/web]
version=3.3.0.Final
#3 (263 bytes):
local_addr=NYC-0/web [b287afd9-70a6-81cd-65ca-561838aaa51f]
cluster=web
RELAY2.printRoutes=LON --> _LON-0/web:LON [UP]
NYC --> _NYC-0/web:NYC [UP]
SFO --> _SFO-0/web:SFO [UP]
physical_addr=127.0.0.1:55400
view=[NYC-0/web|0] [NYC-0/web]
version=3.3.0.Final
#4 (263 bytes):
local_addr=SFO-0/web [b8e9a2ce-72b7-7c14-e420-9276912f827b]
cluster=web
RELAY2.printRoutes=LON --> _LON-0/web:LON [UP]
NYC --> _NYC-0/web:NYC [UP]
SFO --> _SFO-0/web:SFO [UP]
physical_addr=127.0.0.1:55500
view=[SFO-0/web|0] [SFO-0/web]
version=3.3.0.Final
4 responses (4 matches, 0 non matches)
{noformat}
Cross site replication requires totally connected graph of
sites/routes
------------------------------------------------------------------------
Key: WFLY-1416
URL:
https://issues.jboss.org/browse/WFLY-1416
Project: WildFly
Issue Type: Bug
Components: Clustering
Affects Versions: 8.0.0.Alpha2
Reporter: Richard Achmatowicz
Assignee: Paul Ferraro
Cross site replication is configured using JGroups subsystem <relay/> and
<remote-site/> elements to define sites and multicast relay routes between sites.
Looking at a site/route configuration as a graph of nodes and edges, xsite only works if
this graph is totally connected (i.e. every site has a defined direct route to every other
site). This is because <site name> -> <site id> mappings are assigned per
node, with no agreement on what is happening at other nodes.
For example, this configuration works:
<relay site="LON">
<remote-site name="NYC" .../>
<remote-site name="SFO" .../>
</relay>
<relay site="NYC">
<remote-site name="LON" .../>
<remote-site name="SFO" .../>
</relay>
<relay site="SFO">
<remote-site name="NYC" .../>
<remote-site name="LON" .../>
</relay>
but this configuration does not:
<relay site="LON">
<remote-site name="NYC" .../>
<remote-site name="SFO" .../>
</relay>
<relay site="NYC">
<remote-site name="LON" .../>
</relay>
<relay site="SFO">
<remote-site name="LON" .../>
</relay>
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:
http://www.atlassian.com/software/jira