[
https://issues.apache.org/jira/browse/SLING-3316?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Stefan Egli updated SLING-3316:
-------------------------------
Description:
Although discovery.impl can handle situations where a topology connector is
pointing to itself (by correctly ignoring that connector), the connector still
continues pinging every configured heartbeat-interval. The idea behind this
being that the connector should not try to be too smart (KISS) as it could turn
out that eventually that connector no longer points to self but somewhere
useful (eg due to a change in a reverse proxy..)
Nevertheless: turns out that such a local-loop topology-connector can indeed be
a hassle as it fills log files (eg request log, recent requests etc) with
unnecessary information, besides the fact that the pings might indeed not be
necessary.
To improve such a scenario, discovery.impl could stop the connector entirely
when detecting such a local-loop. (The detection is rather trivial: the
response from the PUT connector is already flagged as 'loop=true'. In that case
the connector-client can check if the response came from itself (by verifying
the slingId)).
Since stopping the connector is not something you want in all situations, this
feature should be disabled by default.
was:Although discovery.impl can handle situations where a topology connector
is pointing to itself (by correctly ignoring that connector), the connector
still continues pinging every configured heartbeat-interval. The idea behind
this being that the connector should not try to be too smart (KISS) as it could
turn out that eventually that connector no longer points to self but somewhere
useful (eg due to a change in a reverse proxy..) Nevertheless: turns out that
such a local-loop topology-connector can indeed be a hassle as it fills log
files (eg request log, recent requests etc) with unnecessary information,
besides the fact that the pings might indeed not be necessary. To improve such
a scenario, discovery.impl could stop the connector entirely when detecting
such a local-loop. (The detection is rather trivial: the response from the PUT
connector is already flagged as 'loop=true'. In that case the connector-client
can check if the response came from itself (by verifying the slingId)). Since
stopping the connector is not something you want in all situations, this
feature should be disabled by default.
> Add auto-stop behavior to topology connector if pinging self
> ------------------------------------------------------------
>
> Key: SLING-3316
> URL: https://issues.apache.org/jira/browse/SLING-3316
> Project: Sling
> Issue Type: Improvement
> Components: Extensions
> Affects Versions: Discovery Impl 1.0.2
> Reporter: Stefan Egli
> Assignee: Stefan Egli
>
> Although discovery.impl can handle situations where a topology connector is
> pointing to itself (by correctly ignoring that connector), the connector
> still continues pinging every configured heartbeat-interval. The idea behind
> this being that the connector should not try to be too smart (KISS) as it
> could turn out that eventually that connector no longer points to self but
> somewhere useful (eg due to a change in a reverse proxy..)
> Nevertheless: turns out that such a local-loop topology-connector can indeed
> be a hassle as it fills log files (eg request log, recent requests etc) with
> unnecessary information, besides the fact that the pings might indeed not be
> necessary.
> To improve such a scenario, discovery.impl could stop the connector entirely
> when detecting such a local-loop. (The detection is rather trivial: the
> response from the PUT connector is already flagged as 'loop=true'. In that
> case the connector-client can check if the response came from itself (by
> verifying the slingId)).
> Since stopping the connector is not something you want in all situations,
> this feature should be disabled by default.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
