Repository navigation
Replies: 11 comments 10 replies
|
More information. I noticed that if the file is referenced by the name cache-ispn.xml in keycloak.conf, it is ignored (introducing an error in the XML syntax doesn't disrupt the startup). So I rename it cache-ispn-tcpping.xml. If I set the port in the file like this : This has no effect, Keycloak is listening on default port 7800. If I run this command : Keycloak is listening on port 8888. Thanks, Rémy |
|
I got it working for me after some digging and try&error. In the keycloak.conf I´ve added: Than, I copied cache-ispn.xml to cache-ispn-tcpping.xml, and made it look like this: Afterwards run a build and specify the new cache file, so values get updated. timeout/join_timeout and num_members are not included, because of an error I received: To show that it´s indeed working: |
|
Hi @Nascire It works... I don't know what the default value ss : https://www.keycloak.org/server/caching#_transport_stacks Thanks Rémy |
|
my present keycloak configiruation is to use dns ping as below in wildfly. i just copy pasted the same in my cache-ispn-k8s.yml and could not get my server started. can some one suggest whats the equivalent to be used with quarkus |
|
i would like to use this kind of config in quarkus https://infinispan.org/docs/stable/titles/server/server.html#discovery-dns-ping_cluster-transport and possible pass the dns_query as a runtime/configuration parameter something like kc.[bat|sh] start --dns_query mykeycloakk9sservice-headless.local.cluster I think this will be very useful enhancment to keycloak on quarkus jgroups actually support -D values for query param refer here https://infinispan.org/docs/stable/titles/server/server.html#kubernetes |
|
My two-node cluster, not currently in production, runs on Azure where multicast isn't an option so eventually we went with DNS_PING. I have ... where 10.1.0.1 and 2 are the IP addresses of my two nodes. This appears to work fine. |
|
yes --cache ispn not required for default settings. On Thursday, 28 July, 2022 at 09:21:08 pm IST, Arulaln-AR ***@***.***> wrote:
@eggyknap , thanks. We don't need the --cache ispn right. Because it is the default value. rest all looks good. thanks will try that
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: ***@***.***>
|
That was exactly what I needed @Nascire Thanx! I was so close to this solution already, just didn't know how to override the MPING in the tcp stack. Weird though that UDP multicast is not working (or so hard to get working) on our end. But I used something like this as a workaround before in older keycloak version configurations. |
|
hi, I am a little frustrated here as no matter what I am doing, I cannot make the HA on our swarm, logs below is the same on all the other two nodes. It seems that none of them can see each other (while their container is being created flawlessly), I am using tcpping and mariadb. Any help would be much appreciated.
|
|
how can i test clustering works with tcp ping, can anyone help |
HI, I tried to get TCPPING configured but cannot get it to replace the variables. I start to think that the build is ignoring my xml file, or the varibles in my start.sh file. Here is my cache file. I deleted the original cache-ispn.xml. I am building the config with: And then start it with this script: Yet I cannot get around this replace error : Does anybody know? |
Uh oh!
There was an error while loading. Please reload this page.
Hello,
I managed to do this configuration in the Wildfly distribution. I can't do it in the Quarkus distribution.
I have added this to the file keycloak.conf :
An extract of the cache.xml :
I run this commands :
Port 7800 is open between both servers.
I don't see the discovery of the members in the logs.
The logs are as follows :
When I run this command on the second node :
I've got this :
Thanks,
Rémy
All reactions