Endless build

I try to build a project with four svn repositories and some repository rules to fiddle together the required folder structure to build the project. (I’m re-modelling the svn externals this way).

Unfortunately most of the builds do not end and run forever. It's really strange be cause it behaves differently on each build run.


  • Sometimes the first stage "Build" runs the only action "Visual Studio" successfully. But then the "Agent To Server Workspace Sync" in left in state unknown for ever.
  • Sometimes the “Build” stage completes successfully. But then the “Unit Tests” stage is caught in eternal work. “Server To Agent Workspace Sync” is unknown, the action “MSTest” fails due to not found files and “Agent To Server Workspace Sync” is left in state unknown, too.

Stopping the build manually does not help, either. The build is in state “build stopping” for ever.

Any ideas what's wrong here and what to do to get the builds run and end reliably?


 

Hi Lars,

We’ll need some more details to troubleshoot this issue.

Can you check if there are any errors in the Continua event log at /administration/events?

Which version of Continua are you running? We uploaded a new beta build version 1.0.0.1494 recently with a rewrite of the subversion support to speed up the repository checkout - especially with larger projects. If you haven’t installed that yet please do. You will need to reset your repositories using the Reset buttons on the /administration/ci/repositories page after installing this.

Can you also give us details of your repositories either by attaching screenshots of the Edit Repository dialog or listing the fields for each one? Are these large repositories - can you provide info on their size?

Also can you list the repository and workspace rules for your configuration stages so we can try to replicate your set up?

Hi Dave

thanks for the quick response and sorry for my really incomplete information given.

I'm using the Continua 1.0.0.1494, updated yesterday. I did not reset the repositories but I did it now. While resetting I discovered repositories listed I thought I had deleted earlier (in the project repository configuration). Deletion of repositories seems not reliably working. I hat do delete some of them more than twice. They disappeared after deletion but reappeared a minute later. Finally I got the repositories reflecting what I need.

After that I tried to run a simple project build, which worked fine before the Continua update yesterday (except for the artifact download). Now I discover quite the same behavior as with the complex project. The build runs for ever, unable to stop. Checking the results on the file system, everything was built correctly, even my zip artifacts were synced back to the server. But Continua doesn't get it.Sync state is unknown. The unread error messages state stopped builds and some unreachable repositories during a nightly maintenance of the svn servers. Nothing in there related to this build here. I've attached a screen shot of the build log.

Attachment unavailable

Sorry, the screen shot images are not linked. I attach an archive here.

Ok, I started over with a fresh installation of Continua, because I was afraid I screwed it up somehow. Unfortunately the endless build is still there…

Thanks Lars, we’re looking into it.

Is there a possibility to debug what Continua is donig? Can I help with log files, etc.? It would be a pitty if we had to stay with FB6. Just now I have the possibility to introduce a new CI system here at our company. In my opinion Continua would be a step towards the right direction - if it runs reliably.

Hi Lars

We are in the process of working on this issue, we made some good progress yesterday, hopefully we should have it resolved or at least easier to diagnose today.

In the mean time, can you look in the Windows Eventlog on the machine the agent was running on and see if there are any entries which might help find the cause.

Hi Vencent

Unfortunately the EventLog does not tell too much. I have some errors about “VSTTExecution”. This is related to VisualStudio and testing. PostgeSQL complains about some issues while the system is going down for reboot. Last week I had an access violation exception in Continua.SSHD.exe.

The behaviour is not deterministic. Sometimes the build runs though, sometimes it hangs endlessly in first or second stage. But one thing is very deterministic in case of endless builds: I can’t stop such a build. The state of the build will remain in “Build stopping” until reboot.

Hi Lars

Stopping builds has been the focus of our attention for the last few days, things are looking much better. Hopefully we should have a new build to test in a day or so.

I have been experiencing this same issue for a few days. Just upgraded to 1.0.0.1557 and issue immediately resolved :slight_smile: Tx

After upgrading to 1557, I can get builds which fail correctly, but then get stuck on ‘Agent To Server Workspace Sync’ after the failure. For example, I have one stage which failed a FinalBuilder action after 15 seconds, but has been stuck on ‘Agent To Server Workspace Sync’ for 40 minutes. Attempting to manually stop the build via the ‘stop build’ button doesn’t help either. I’m left with no choice but to restart Continua.

What’s frustrating is it happens intermittently.

Hi Joel

We’re still working on this… it’s proving very tough to find the cause.

Hi Vincent,
I updated to v1.0.0.1557 but unfortunately this does not fix the issue. No change at all. The event log onl states a not reachable svn repository in the night hours (probably due to maintenance of the svn servers )
Lars

Hi Lars

The unreachable repository at night can be resolved but specifying downtime on the repository.

With regards to the endless build, can you check if the agent is using the share or SSH to transfer files to the server. To do that, Admin menu, CI Server, where the agent lists how many properties , click on that link, and check the values of

serverfiletransport.sshavailable

and

serverfiletransport.uncavailable

The next thing to do, is edit the configuration, go to the stages page, double click on the stage, go to the Workspace Rules tab and check the “Log workspace files copied” option. That will write the list of files copied to the log.

We are still working on this issue, we have not been able to reproduce the problem (endless stopping we have), however we have been testing the error handling by injecting errors all over the place and have found and fixed lots of issues. Hopefully that will mean whatever error is happening on your system will at least show up somewhere we can see it.

Hi Vincent,
here is our configuration:
[code]serverfiletransport.forcessh Falseserverfiletransport.sshavailableTrueserverfiletransport.uncavailableTrue [/code]

Attachment unavailable