[
https://issues.apache.org/jira/browse/SLING-12436?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Christian Schneider updated SLING-12436:
----------------------------------------
Description:
org.apache.sling.distribution.journal.impl.discovery.DiscoveryService is
configured to be called by sling scheduler every 5 seconds.
{noformat}
// Register periodic task to update the topology view
Dictionary<String, Object> props = new Hashtable<>();
props.put(PROPERTY_SCHEDULER_CONCURRENT, false);
props.put(PROPERTY_SCHEDULER_PERIOD, 5L); // every 5 seconds
reg = context.registerService(Runnable.class.getName(), this, props);{noformat}
This makes sling scheduler use the default thread pool. At a customer we had
some other job block all threads of default thread pool. This caused our code
here to not be called anymore and thus the UI of content distribution showed no
progress in replication queues. For customers this looked like replication
stopped working.
We should set a specific thread pool for distribution to make sure we are not
affected by other code using sling scheduler.
Apart from this we should also setup the same for
ResourceDistributionPackageCleanup
h3.
was:
org.apache.sling.distribution.journal.impl.discovery.DiscoveryService is
configured to be called by sling scheduler every 5 seconds.
{code:java}
// Register periodic task to update the topology view Dictionary<String,
Object> props = new Hashtable<>();
props.put(PROPERTY_SCHEDULER_CONCURRENT, false);
props.put(PROPERTY_SCHEDULER_PERIOD, 5L); // every 5 seconds
reg = context.registerService(Runnable.class.getName(), this, props); {code}
This makes sling scheduler use the default thread pool. At a customer we had
some other job block all threads of default thread pool. This caused our code
here to not be called anymore and thus the UI of content distribution showed no
progress in replication queues. For customers this looked like replication
stopped working.
We should set a specific thread pool for distribution to make sure we are not
affected by other code using sling scheduler.
Apart from this we should also setup the same for
ResourceDistributionPackageCleanup
h3.
> Change scheduler setup in distribution journal to be more resilient
> -------------------------------------------------------------------
>
> Key: SLING-12436
> URL: https://issues.apache.org/jira/browse/SLING-12436
> Project: Sling
> Issue Type: Improvement
> Reporter: Christian Schneider
> Priority: Major
>
> org.apache.sling.distribution.journal.impl.discovery.DiscoveryService is
> configured to be called by sling scheduler every 5 seconds.
>
> {noformat}
> // Register periodic task to update the topology view
> Dictionary<String, Object> props = new Hashtable<>();
> props.put(PROPERTY_SCHEDULER_CONCURRENT, false);
> props.put(PROPERTY_SCHEDULER_PERIOD, 5L); // every 5 seconds
> reg = context.registerService(Runnable.class.getName(), this,
> props);{noformat}
>
> This makes sling scheduler use the default thread pool. At a customer we had
> some other job block all threads of default thread pool. This caused our code
> here to not be called anymore and thus the UI of content distribution showed
> no progress in replication queues. For customers this looked like replication
> stopped working.
> We should set a specific thread pool for distribution to make sure we are not
> affected by other code using sling scheduler.
>
> Apart from this we should also setup the same for
> ResourceDistributionPackageCleanup
> h3.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
