Uploaded image for project: 'mod_cluster'
  1. mod_cluster
  2. MODCLUSTER-759

watchdog thread load increases with balancer members

XMLWordPrintable

      The mod_cluster watchdog thread currently invokes operations every 1 second. It currently iterates over all members 3 times, once in each of these calls:

                  /* removed nodes: check for workers */
            remove_workers_nodes(conf, pool, s);
            /* cleanup removed node in shared memory */
            remove_removed_node(pool, s);
            /* Calculate the lbstatus for each node */
            update_workers_lbstatus(conf, pool, s);

      So this can result in some growing CPU load as JVMs increase. We may consider decreasing this complexity or frequency as possible.

              rhn-engineering-jclere Jean-Frederic Clere
              rhn-support-aogburn Aaron Ogburn
              Votes:
              1 Vote for this issue
              Watchers:
              3 Start watching this issue

                Created:
                Updated:
                Resolved: