Skip to content

restore checkpoint app image start slower when using sessionCache with infinispan #552

Description

@tam512

Restore eBuy checkpoint app image on OCP. The app is configured to have httpSessionCache with infinispan with the following in server.xml

<server>
    <featureManager>
	    <feature>appSecurity-4.0</feature>
	    <feature>sessionCache-1.0</feature> 
    </featureManager>

    <library id="DatagridLib">
	    <fileset dir="${server.config.dir}/datagrid" includes="*.jar"/>
    </library>

    <cacheManager id="CacheManager" uri="file:///${server.config.dir}/datagrid/infinispan_hotrod.props">
	    <properties infinispan.client.hotrod.uri="hotrod://${INFINISPAN_USER}:${INFINISPAN_PASS}@${INFINISPAN_HOST}:${INFINISPAN_PORT}"/>
        <cachingProvider providerClass="org.infinispan.jcache.remote.JCachingProvider" jCacheLibraryRef="DatagridLib" />
    </cacheManager>

    <httpSessionCache cacheManagerRef="CacheManager" cacheSeparator="_"/>

</server>

When restore eBuy app on OCP without using httpSessionCache, it takes less than 1 second or sometimes just a little over 1 second for the server to start, but when using httpSessionCache, it takes about 12 seconds for the first app pod to start and about 4 to 5 seconds for the later pods

I see that there are many org.infinispan.HOTROD messages in the logs that add more time to the server start time

pods' logs: pods-logs.zip

Slack discussion: https://ibm-cloud.slack.com/archives/C03MR7EC3NG/p1718041901887979?thread_ts=1717187652.885869&cid=C03MR7EC3NG and https://ibm-cloud.slack.com/archives/C03MR7EC3NG/p1718226721233629

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions