Collaborative Editing fails on clustered Confluence Data Center - can't establish a persistent WebSocket connection


アトラシアン コミュニティをご利用ください。



プラットフォームについて: Data Center のみ - この記事は、Data Center プラットフォームのアトラシアン製品にのみ適用されます。

この KB は Data Center バージョンの製品用に作成されています。Data Center 固有ではない機能の Data Center KB は、製品のサーバー バージョンでも動作する可能性はありますが、テストは行われていません。サーバー*製品のサポートは 2024 年 2 月 15 日に終了しました。サーバー製品を利用している場合は、アトラシアンのサーバー製品のサポート終了のお知らせページにて移行オプションをご確認ください。

*Fisheye および Crucible は除く


After setting up a Clustered Confluence Data Center to run behind an Amazon Web Service Elastic Load Balancer, Collaborative Editing isn't working.

The error message displayed is The editor didn't load this time, and you already reviewed the document Confluence throws "The editor didn't load this time" error when trying to edit a page.


Clustered Confluence Data Center using Self-Managed Synchrony

ロード バランサ


  1. When capturing a HAR file, while editing a document with Collaborative Editing enabled, it's possible to see the WebSocket connection upgrade (HTTP 101 - Switching protocols). The WebSocket connection is a persistent connection and should be active during all the time the document is being edited. However, when we can see multiple requests for the Websocket upgrade, all taking a long time to complete, as for example:

  2. In Synchrony logs of Node 1, we can see a ConnectTimeoutException happening when this node tries to communicate with Node 2 using Port 25500:

    atlassian-synchrony.log for Node 1
    2020-11-17 15:41:07,626 WARN [async-dispatch-12] [synchrony.event-bus] error creating topic connection {:throwable #error {
    2020-11-17 15:41:07,626 DEBUG [2571:StdOutHandler [/data/atlassian/confluence/jre/bin/java]] 	 :cause "connection timed out: /<NODE2_IP_ADDRESS>:25500"
    2020-11-17 15:41:07,626 DEBUG [2571:StdOutHandler [/data/atlassian/confluence/jre/bin/java]] 	 :via
    2020-11-17 15:41:07,626 DEBUG [2571:StdOutHandler [/data/atlassian/confluence/jre/bin/java]] 	 [{:type
    2020-11-17 15:41:07,626 DEBUG [2571:StdOutHandler [/data/atlassian/confluence/jre/bin/java]] 	   :message "connection timed out: /<NODE2_IP_ADDRESS>:25500"
  3. On Node 2, we can see that Node 1 was reported dead, and the Cluster Membership is composed of Node 2 only:

    atlassian-synchrony.log for Node 1
    2020-11-17 15:41:34,724 WARN [hz._hzInstance_1_cluster-name-Synchrony.cached.thread-3] [internal.cluster.impl.MembershipManager] [<NODE2_IP_ADDRESS>]:5701 [cluster-name-Synchrony] [3.11.4] Member [<NODE1_IP_ADDRESS]:5701 - 610548f6-f067-4a9d-9a73-7e5c5a9b8491 is suspected to be dead for reason: No connection
  4. When trying to test the WebSocket upgrade through curl, using Confluence Base URL, we receive an empty reply from the server instead of establishing a WebSocket connection:

    WebSocket upgrade test
    $ curl -i -N -H "Connection: Upgrade" -H "Upgrade: websocket" -H "Sec-WebSocket-Key: 33xyqDvzAXTYgsjjbaYD5A==" --header "Sec-WebSocket-Version: 13" https://<CONFLUENCE_BASE_URL>/synchrony/v1/bayeux-sync1
      % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                     Dload  Upload   Total   Spent    Left  Speed
      0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0HTTP/1.1 101 Switching Protocols
    Date: Thu, 19 Nov 2020 02:05:57 GMT
    Connection: upgrade
    P3P: CP="This is not a P3P policy! See for more info."
    upgrade: websocket
    sec-websocket-accept: RV68WHXjNM0H+IFE5/W1ioLmMHI=
      0     0    0     0    0     0      0      0 --:--:--  0:01:00 --:--:--     0
    curl: (52) Empty reply from server


The inter-node communication in Synchrony isn't working properly using the ports mentioned on the log files.


  • Review the setup of all ports used by Synchrony on nodes of a Confluence Data Center setup. Ensure that:
    • All nodes are listening on ports 5701 (Hazelcast for Synchrony) and 25500 (Cluster base port for Synchrony).
    • You can establish a connection from one node to the other using these ports

最終更新日: 2020 年 12 月 1 日


Powered by Confluence and Scroll Viewport.