Uploaded image for project: 'Couchbase Server'
  1. Couchbase Server
  2. MB-7196

moxi doesn't seem to failover on downstream hung

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: Major
    • Resolution: Won't Do
    • Affects Version/s: 1.8.1
    • Fix Version/s: None
    • Component/s: moxi
    • Security Level: Public
    • Labels:
      None
    • Environment:
    • Triage:
      Untriaged

      Description

      just run a couple of nodes, freeze one of the downstreams with a "kill -STOP"
      while running this simple piece of code:

      <?php
       
      ini_set('display_errors', 1);
      error_reporting(-1);
       
      if (! extension_loaded('memcached')) {
          dl('memcached.so');
      }
      $sessionId = $argc > 1 ? $argv[1] : md5(mt_rand());
       
      ini_set('session.save_handler', 'memcached');
      ini_set('session.save_path', 'localhost:11211'); // host:port
       
      session_id($sessionId);
      session_start();
       
      printf("Session %s: %s\n", $sessionId, var_export($_SESSION, true));
       
      $_SESSION['timestamp'] = time();
      

      it will randomly fail

      p.s. initially mentioned at MB-3023

      No reviews matched the request. Check your Options in the drop-down menu of this sections header.

        Activity

        Hide
        zeph Guido Serra added a comment -

        v1:CouchBase zeph$ php memcache-session-1.php
        Session c6edc9f4624f78cdfa49af134d60887b: array (
        )
        PHP Warning: Unknown: Failed to write session data (memcached). Please verify that the current setting of session.save_path is correct (localhost:11211) in Unknown on line 0

        Warning: Unknown: Failed to write session data (memcached). Please verify that the current setting of session.save_path is correct (localhost:11211) in Unknown on line 0

        Show
        zeph Guido Serra added a comment - v1:CouchBase zeph$ php memcache-session-1.php Session c6edc9f4624f78cdfa49af134d60887b: array ( ) PHP Warning: Unknown: Failed to write session data (memcached). Please verify that the current setting of session.save_path is correct (localhost:11211) in Unknown on line 0 Warning: Unknown: Failed to write session data (memcached). Please verify that the current setting of session.save_path is correct (localhost:11211) in Unknown on line 0
        Hide
        cwt cwt added a comment -

        I am seeing this same issue in Couchbase-server 2.2.0. Is there any new development for this issue?

        Show
        cwt cwt added a comment - I am seeing this same issue in Couchbase-server 2.2.0. Is there any new development for this issue?
        Hide
        drigby Dave Rigby added a comment -

        It's not clear to me what the OP is expecting to happen here. Moxi has no concept of "failover" - it's not an active clustering system which somehow could failover to a different node if it experiences issues communicating.

        In effect moxi is just another "Smart client" - it knows about the Couchbase cluster topology and where a given vBucket is active, but in the event a smart client cannot communicate to one of the cluster nodes (e.g. simulated network issue by sending SIGSTOP to a node) then it will simply report an error to the user (possibly after a certain timeout).

        Given how long this issue has been open, the fact that IMHO there is no bug here and that moxi is essentially in maintenance-only mode I'm closing this as "Won't Do". If you disagree then please re-open with additional information on how the expected behaviour differs from the actual behaviour you are seeing.

        Show
        drigby Dave Rigby added a comment - It's not clear to me what the OP is expecting to happen here. Moxi has no concept of "failover" - it's not an active clustering system which somehow could failover to a different node if it experiences issues communicating. In effect moxi is just another "Smart client" - it knows about the Couchbase cluster topology and where a given vBucket is active, but in the event a smart client cannot communicate to one of the cluster nodes (e.g. simulated network issue by sending SIGSTOP to a node) then it will simply report an error to the user (possibly after a certain timeout). Given how long this issue has been open, the fact that IMHO there is no bug here and that moxi is essentially in maintenance-only mode I'm closing this as "Won't Do". If you disagree then please re-open with additional information on how the expected behaviour differs from the actual behaviour you are seeing.

          People

          • Assignee:
            drigby Dave Rigby
            Reporter:
            zeph Guido Serra
          • Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved:

              Gerrit Reviews

              There are no open Gerrit changes