In the past week (since the upgrade to 8.1.4, may or may not be relevant) we have had odd behaviour when deploying changes between environments on Umbraco cloud.
Deploying a change from Development to Live is successful, but the public website appears to have all the content unpublished. Looking at nodes in the back office, they will say "This node is published but not in the cache", or "This document is published but its url cannot be routed".
Pressing Save and Republish on an affected node throws an error the first time, but will work on the second try....
Unhandled controller exception occurred for request
'"https://XXX.s1.umbraco.io/umbraco/backoffice/UmbracoApi/Content/PostSave"'
Exception System.Exception: panic: failed to get next child with id=-1
at Umbraco.Web.PublishedCache.NuCache.ContentStore.GetLinkedNode(Int32
id, String description) at
Umbraco.Web.PublishedCache.NuCache.ContentStore.RemoveNodeLocked(ContentNode
content) at
Umbraco.Web.PublishedCache.NuCache.ContentStore.SetBranch(Int32
rootContentId, IEnumerable'1 kits) at
Umbraco.Web.PublishedCache.NuCache.PublishedSnapshotService.NotifyLocked(IEnumerable'1
payloads, Boolean& draftChanged, Boolean& publishedChanged) at
Umbraco.Web.PublishedCache.NuCache.PublishedSnapshotService.Notify(JsonPayload[]
payloads, Boolean& draftChanged, Boolean& publishedChanged) at
Umbraco.Web.Cache.ContentCacheRefresher.Refresh(JsonPayload[]
payloads) at
Umbraco.Core.Sync.ServerMessengerBase.DeliverLocal[TPayload](ICacheRefresher
refresher, TPayload[] payload)
There are no errors in the deployment log, but this problem has happened consistently with the past few deployments we have made to Live and I think on development too (but harder to spot).
The easiest fix is to go into Settings > Published Status and press Reload on the Memory cache. So it doesn't look like it's causing any real harm, but my biggest concern is that an automatic upgrade at a random time on a Tuesday will unpublish a production site without warning ....
This site was impacted by having inconsistent language settings (trying to change en-us to en-gb) which caused deployment issues I think back on 8.1.2 but I thought that was dealt with.
Has anyone seen anything similar on their sites? Or have clues as to where to look further to investigate?
We had a similar issue although not on cloud, however we are using models builder and hydrating extended models with data from external source and was causing an issue see
It still happens, but I think only when the Document Types are changed, and presumably Deploy has more work to do recreating caches. Fortunately the automatic upgrades haven't knocked out the site (so far) - probably for this reason.
Tweaks to views or macros don't cause a problem. And my earlier suspicion of mismatched language settings is probably a red herring.
It may be well be related to locks in nucache - but I don't know if that is a symptom or a cause.
It did it again for me this morning - simply making a field not mandatory was enough to "unpublish" all documents of that type when deploying to live. Refreshing memory cache fixed it. No errors in the log or in the Kudu deployment logs :(
Deploying between cloud environments empties the memory cache (unpublishing the site!)
Hello,
In the past week (since the upgrade to 8.1.4, may or may not be relevant) we have had odd behaviour when deploying changes between environments on Umbraco cloud.
Deploying a change from Development to Live is successful, but the public website appears to have all the content unpublished. Looking at nodes in the back office, they will say "This node is published but not in the cache", or "This document is published but its url cannot be routed".
Pressing Save and Republish on an affected node throws an error the first time, but will work on the second try....
There are no errors in the deployment log, but this problem has happened consistently with the past few deployments we have made to Live and I think on development too (but harder to spot).
The easiest fix is to go into Settings > Published Status and press Reload on the Memory cache. So it doesn't look like it's causing any real harm, but my biggest concern is that an automatic upgrade at a random time on a Tuesday will unpublish a production site without warning ....
This site was impacted by having inconsistent language settings (trying to change en-us to en-gb) which caused deployment issues I think back on 8.1.2 but I thought that was dealt with.
Has anyone seen anything similar on their sites? Or have clues as to where to look further to investigate?
Hey, I have a same issue on a Cloud 8.2.
Have you found a solution / fix ?
We had a similar issue although not on cloud, however we are using models builder and hydrating extended models with data from external source and was causing an issue see
https://our.umbraco.com/forum/using-umbraco-and-getting-started/99670-this-document-is-published-but-is-not-in-the-cache-after-build
There is also potentially issue of locks no nucache after a deploy so it could be that?
Hi, we have the same problem on all 8.2.0 Cloud Projects and we do not use Model Builder.
I hope it's fixed in 8.2.1 that are shipping on tuesday, else the automatic upgrade will lay all sites down :)
It still happens, but I think only when the Document Types are changed, and presumably Deploy has more work to do recreating caches. Fortunately the automatic upgrades haven't knocked out the site (so far) - probably for this reason.
Tweaks to views or macros don't cause a problem. And my earlier suspicion of mismatched language settings is probably a red herring.
It may be well be related to locks in nucache - but I don't know if that is a symptom or a cause.
It did it again for me this morning - simply making a field not mandatory was enough to "unpublish" all documents of that type when deploying to live. Refreshing memory cache fixed it. No errors in the log or in the Kudu deployment logs :(
To anyone following this thread, I'm hopeful that a fix to nucache that was in 8.2.2 may have fixed this.
https://github.com/umbraco/Umbraco-CMS/pull/7002
But I've not done a schema change / release to find out yet. Does anyone else have reports of a fix (or not) since the release on Tuesday?
I havn't had the problem since 8.2.2.. Yet :)
is working on a reply...