Details
Description
In cluster, having two servers and updating some object on server1, the update is only seen on server2 after clearing caches.
Steps to Reproduce
Having two servers configured in cluster
(Server1)
- Create one folder and name it "Folder1"
(Server2) - View that "Folder1" is present (it is)
(Server1) - Update "Folder1" renaming it to "Folder 1.1"
(Server2) - Call back server2 again and try to view the folder again. Here we can see that is still named "Folder1". This step might be done using either Dashboard or content tree
After clearing caches on server2, then we can see that the update is now seen.
Trying again without first seeing "Folder1" on server2 (and thus, without generating cache)
(Server1)
- Create one folder and name it "Folder1"
(Don't see it on server2) - Update "Folder1" renaming it to "Folder 1.1"
(Server2) - Try to view the folder in server2. Now we can see that the name is correct: "Folder 1.1"
--------------------------------
My configuration:
I have 4 servers running, where
Server1: 192.168.2.221: (ezdfs1.paulo) Where one of the ezplatform installations is
Server1: 192.168.2.221: (ezdfs2.paulo) Where another ezplatform installations is
Server3: 10.0.5.4: Where the database is. Both server1 and server2 share this database
Server4: 192.168.2.201: My varnish server.
I set my php.ini in both server1 and server2 with
session.save_handler = memcached
session.save_path = "ezdfs1.paulo:11211"
On varnish, I configured with round_robin, but for better test this issue, one can configure it with "hash" and assign different priorities to each server.
My ezplatform.yml, vhost and varnish vcl files are attached to this issue.
Attachments
Issue Links
- testing discovered
-
EZP-25972 Clustering guide must explain shared stash cache / sessions better
- Closed