Bottleneck when using auth/admin/realms/myrealm/users in my app

Question

I am running a script which requests the master token before requesting the list of users for a realm. I perform this request in a tight loop 100 times and output the time it takes to perform the "auth/admin/realms/mine/users?first=0&max=1000" request.

When testing with a locally installed keycloak docker, a single request is taking 4.0 seconds. If I run 5 instances of this script at the same time, a single request takes around 10 seconds. And if I run 10 instances, it jumps to 20 seconds before a reply is received.

Therefore, I have a serious bottleneck in my webapp on the page where I list all user information and was wondering how I might go about solving this problem.

I am running keycloak 2.5.0.Final. My java home is /usr/lib/jvm/java-1.8.0-openjdk-1.8.0.111-2.b15.el7_3.x86_64/jre. authorizationPersister = jpa.

My script:

import requests

keyCloakUrl = "http://foo.bar"

for i in range(100):

    session = requests.Session()
    r = session.post(
        keyCloakUrl +"auth/realms/master/protocol/openid-connect/token", 
        data={
            "grant_type":"password",
            "client_id":"admin-cli",
            "username":"admin",
            "password":"admin"
        }
    )
    master_token = json.loads(r.content)['access_token']

    r = s.get(
        keyCloakUrl + "auth/admin/realms/mine/users?first=0&max=1000",
        headers={'Authorization':'Bearer '+master_token}
    )
    print(r.elapsed.total_seconds())

My dockerfile:

FROM jboss/keycloak-mysql:2.5.0.Final

ARG db2

ADD deps/ /opt/jboss/keycloak/
RUN /opt/jboss/keycloak/import_utbud_realm.sh 

CMD ["-b", "0.0.0.0", "-bmanagement", "0.0.0.0", "-Djboss.socket.binding.port-offset=2"]

UPDATE

I also tried measuring the time for the "auth/realms/master/protocol/openid-connect/token" request. This is 0.1 seconds per request with only one instance of the script running and 0.8 seconds with 10 instances running.

Answer 1

Maybe it is a silly observation, but did you try to increase you JVM heap in your docker configs? Like:

CMD java -XX:+PrintFlagsFinal -XX:+PrintGCDetails $JAVA_OPTIONS -jar java-container.jar

There is a fantastic blog that explain the memory issue with the JVM and docker here