Umbraco Cloud - Manually trigger build and deploy without git check in.
I get errors fairly often in Umbraco cloud on git pushes related to the build and deploy process. They are not related to the push itself. In order to get the deploy to occur, I am pushing inconsequential commits (like adding a space to a code file); until I get the build and deploy to complete. There has to be a better way, but I cannot find it.
Is there a way from the cloud console to just say "Run Build with Current Code" and "Run Deploy with Most recent Build"?
Example of a common failure:
git -c diff.mnemonicprefix=false -c core.quotepath=false
--no-optional-locks push -v origin master:master warning: auto-detection of host provider took too long (>2000ms) warning: see
https://aka.ms/gcm/autodetect for more information. warning:
auto-detection of host provider took too long (>2000ms) warning: see
https://aka.ms/gcm/autodetect for more information. POST
git-receive-pack (830 bytes) remote: Updating branch 'master'.
remote: Updating submodules. remote: Preparing deployment for
commit id '6048722b09'. remote: Running custom deployment
command... remote: Running deployment command...
remote: Ensuring git configuration remote: Handling ASP.NET
Core Web Application deployment remote: CsProjFile location
C:\home\site\repository\src\SIAM.Site\SIAM.Site.csproj remote:
Target Major and Minor dotnet sdk version found via the csproj file
'8.0' remote: Working Sdk version '8.0.205' remote:
Wrote global.json in the repository for version '8.0.205'
remote: Determining projects to restore... remote:
...................... send-pack: unexpected disconnect while reading
sideband packet error: error in sideband demultiplexer
I was hoping for something in the umbraco cloud user interface, but I do not think it exists. Marking this as answer as I think this is likely the best answer.
I have had this issue ongoing for months and it is driving me insane. I am working with support again today on this. I too am on useast01. I often need to do up to 4 "touch" pushes from local to dev just to get the package to deploy to remote dev. I am convinced it has to do with an overburdened API - maybe too many automated CI/CD processes, who knows.
It is definitely not a networking issue between local and cloud because the changes appear in the cloud back office, it's just that the deployment fails from there.
I agree that deploy needs a way to deal with this situation via the deploy backoffice UI with some buttons similar to the other deploy functions such as "Clear Signatures" etc.
@Soren, can you share a sample PUT request with the format of the auth headers? That would be great!
Thanks for posting this to confirm that it's not just me...
If anyone else comes here with the "send-pack: unexpected disconnect while reading sideband packet error: error in sideband demultiplexer" issue, I was able to get this resolved with the assistance of Umbraco Support.
I was directed to delete the C:\home\site\wwwroot\umbraco\Data\TEMP folder using Kudo on my Development instance.
Once that was done, I executed a push from my local to the cloud repo which kicked off the deployment process.
We were continually getting this error on shared. Eventually, this prompted us to move to dedicated. That fixed this error, however we then encountered a rash of other issues when then pushing from remote dev to production - 1+ hour deploy times, multiple production reboots, etc. Support was great working with us throughout this process, but were not ever able to fully resolve the issues with deploy.
Eventually (after 6+ weeks), we decided to create a brand new "fresh" project back on shared, restored the database into the project and then migrated in all our projects and files similar to what you would do if you were moving regions: https://docs.umbraco.com/umbraco-cloud/getting-started/migrate-between-regions.
Needless to say it was a lot of work to reconfigure a complicated project, but blowing away 2 years of git history seemed to have done the trick for now. We are on week two with no issues. I will post back here if things change.
We have a more atypical set up and would be interested to hear if anyone is using the following features. Maybe this can help narrow things down.
3 projects in the solution (core, models builder models, and web)
Using config (web.config) transforms for 301 redirection
Renamed the main umbraco project to a custom name
Models builder is enabled on local dev only and is off on remote dev and prod
I have various unproven theories on this error, but I think it's related to network traffic/resourcing on the API that orchestrates all the deployments. Since the packages are landing on the cloud UI dashboard, you know the changes have made it to their destination, it's just that deploy does not finish the job. Unfortunately this git error is really cryptic and can mean a bunch of different things. That said, it seems to happen mostly during prime business hours in the local region, with less or no occurrences after hours. Unfortunately I don't have enough knowledge of the architecture to really make an educated guess, but this is based on my observations to date.
Umbraco Cloud - Manually trigger build and deploy without git check in.
I get errors fairly often in Umbraco cloud on git pushes related to the build and deploy process. They are not related to the push itself. In order to get the deploy to occur, I am pushing inconsequential commits (like adding a space to a code file); until I get the build and deploy to complete. There has to be a better way, but I cannot find it.
Is there a way from the cloud console to just say "Run Build with Current Code" and "Run Deploy with Most recent Build"?
Example of a common failure:
Hi Jonathan
You can send a PUT request to https://your-project-alias.scm.useast01.umbraco.io/api/deployments (you need to add basic auth headers with your Umbraco ID credentials). This will trigger the deployment from the current status in git.
I was hoping for something in the umbraco cloud user interface, but I do not think it exists. Marking this as answer as I think this is likely the best answer.
The associated documentation can be found here: https://docs.umbraco.com/umbraco-cloud/set-up/project-settings/umbraco-cicd/umbracocloudapi
Thank you.
I have had this issue ongoing for months and it is driving me insane. I am working with support again today on this. I too am on useast01. I often need to do up to 4 "touch" pushes from local to dev just to get the package to deploy to remote dev. I am convinced it has to do with an overburdened API - maybe too many automated CI/CD processes, who knows.
It is definitely not a networking issue between local and cloud because the changes appear in the cloud back office, it's just that the deployment fails from there.
I agree that deploy needs a way to deal with this situation via the deploy backoffice UI with some buttons similar to the other deploy functions such as "Clear Signatures" etc.
@Soren, can you share a sample PUT request with the format of the auth headers? That would be great!
Thanks for posting this to confirm that it's not just me...
We have also been struggling with this error for several months.
We are on euwest01.
Sometimes 3 days go by without any problems, but then suddenly this error occurs and we can't deploy for hours. That's nerve-wracking.
In my opinion, triggering the deployment manually would only be a temporary woraround.
I would be more interested to know where the actual "error in sideband demultiplexer" issue comes from and how to fix it.
I'm glad to hear, that we're apparently not the only ones with this problem...
Another observation we were able to make - for us the error almost always occurs on Development, very rarely on Staging, and never on Live.
It seems to us that the error is a result of frequent deployments, which would indicate an overburdened API, like @Jamie mentioned.
If anyone else comes here with the "send-pack: unexpected disconnect while reading sideband packet error: error in sideband demultiplexer" issue, I was able to get this resolved with the assistance of Umbraco Support.
I was directed to delete the C:\home\site\wwwroot\umbraco\Data\TEMP folder using Kudo on my Development instance.
Once that was done, I executed a push from my local to the cloud repo which kicked off the deployment process.
This time, the deployment completed successfully.
@Steve Adams
We are working on that together with the Umbraco Support since September, but we still have no proper solution.
I tried to delete this folder, but it didn't fix the issue for us.
The only thing that worked for a short period of time was to switch to dedicated ressources, and back again. But after a week we got the error again.
We were continually getting this error on shared. Eventually, this prompted us to move to dedicated. That fixed this error, however we then encountered a rash of other issues when then pushing from remote dev to production - 1+ hour deploy times, multiple production reboots, etc. Support was great working with us throughout this process, but were not ever able to fully resolve the issues with deploy.
Eventually (after 6+ weeks), we decided to create a brand new "fresh" project back on shared, restored the database into the project and then migrated in all our projects and files similar to what you would do if you were moving regions: https://docs.umbraco.com/umbraco-cloud/getting-started/migrate-between-regions.
Needless to say it was a lot of work to reconfigure a complicated project, but blowing away 2 years of git history seemed to have done the trick for now. We are on week two with no issues. I will post back here if things change.
We have a more atypical set up and would be interested to hear if anyone is using the following features. Maybe this can help narrow things down.
I have various unproven theories on this error, but I think it's related to network traffic/resourcing on the API that orchestrates all the deployments. Since the packages are landing on the cloud UI dashboard, you know the changes have made it to their destination, it's just that deploy does not finish the job. Unfortunately this git error is really cryptic and can mean a bunch of different things. That said, it seems to happen mostly during prime business hours in the local region, with less or no occurrences after hours. Unfortunately I don't have enough knowledge of the architecture to really make an educated guess, but this is based on my observations to date.
is working on a reply...