We have identified an issue which I hope someone can help with. We have a load balanced setup using flexible load balancing with a single master server. I have tested this using vanilla 7.6.3, 7.6.11 and 7.7.4 with the same results in each (no packages or modifications other than the starter kit that comes with these builds)
We create a page with sub-pages. If we make a change to a sub page and save the change (not publish the change), then un-publish and re-publish the parent page, the original published version of the sub-page will work on the master server but not on the FE servers.
The only way we have found to allow the original published version of the sub-page (with a saved change) on the FE servers is to delete the umbraco.config and restart the application pool.
Anyone have any ideas if this is an issue or by design? We really need to find a way to have the original published version of the sub-page without deleting the umbraco.config and restarting the app pool.
We are running on vmware VM's with Windows 2012 R2 servers (patches up to date). I'm actually using a test lab setup to replicate a Live server issue.
One point I forgot to mention is that we are using DFS-R for replication, but excluding the Temp folder and umbraco.config file, as per the Option 2 instructions.
So my understanding (please correct me if I am wrong) is that distributed calls was the old way of load balancing prior to 7.3 which used a Master server (the first server in the list) to call the other servers when the cache needed updating - Push method I believe?
From 7.3, I believe that each front end node checks a cache table in the database therefore there is no requirement anymore to keep track of which servers are in the load balance cluster as they contact the db directly.
Saying that, I did try enabling distributed calls and added the list of servers, but that did not make any difference.
So I have found Republish entire site fixes the issue rather than deleting the umbraco.config on the front end nodes, as republish entire site does not "republish" as in publish, but refreshes the cache.
I think this is still an issue where the sub-page is not republished correctly on the frond end nodes.
Publishing issue with load balanced servers
Hi,
We have identified an issue which I hope someone can help with. We have a load balanced setup using flexible load balancing with a single master server. I have tested this using vanilla 7.6.3, 7.6.11 and 7.7.4 with the same results in each (no packages or modifications other than the starter kit that comes with these builds)
We create a page with sub-pages. If we make a change to a sub page and save the change (not publish the change), then un-publish and re-publish the parent page, the original published version of the sub-page will work on the master server but not on the FE servers.
The only way we have found to allow the original published version of the sub-page (with a saved change) on the FE servers is to delete the umbraco.config and restart the application pool.
Anyone have any ideas if this is an issue or by design? We really need to find a way to have the original published version of the sub-page without deleting the umbraco.config and restarting the app pool.
Thanks
Hi Alex,
Are you running in Azure by any chance?
Also, have you taken a look at the following doc: https://our.umbraco.org/documentation/getting-started/setup/server-setup/load-balancing/flexible#umbraco-xml-cache-file
Regards
Craig
Hi Craig,
No, not in Azure. We are using "Option #2 : File Storage with File Replication" - https://our.umbraco.org/documentation/getting-started/setup/server-setup/load-balancing/files-replicated
We are running on vmware VM's with Windows 2012 R2 servers (patches up to date). I'm actually using a test lab setup to replicate a Live server issue.
One point I forgot to mention is that we are using DFS-R for replication, but excluding the Temp folder and umbraco.config file, as per the Option 2 instructions.
Thanks
Hi Alex,
Did you add all your servers to umbracoSettings.config ?
https://our.umbraco.org/documentation/Reference/Config/umbracoSettings/#distributedcalls-loadbalancing
Dave
Hi Dave,
So my understanding (please correct me if I am wrong) is that distributed calls was the old way of load balancing prior to 7.3 which used a Master server (the first server in the list) to call the other servers when the cache needed updating - Push method I believe?
From 7.3, I believe that each front end node checks a cache table in the database therefore there is no requirement anymore to keep track of which servers are in the load balance cluster as they contact the db directly.
Saying that, I did try enabling distributed calls and added the list of servers, but that did not make any difference.
Alex
Hi,
So I have found Republish entire site fixes the issue rather than deleting the umbraco.config on the front end nodes, as republish entire site does not "republish" as in publish, but refreshes the cache.
I think this is still an issue where the sub-page is not republished correctly on the frond end nodes.
is working on a reply...