We have a situation where our builds have stopped executing in a stable manner.
At a rate of about one every three we receive either TF215096 or TF215097 errors & the Build fails.
If we then restart the Build controller, it works again – until next time.
The errors we get are:
TF215096: An error occurred while connecting to controller vstfs:///Build/Controller/1: There was no endpoint listening at ht*p://XXXX that could accept the message. This is often caused by an incorrect address or SOAP action. See InnerException, if present, for more details.
TF215096: An error occurred while connecting to controller XXX – Controller: Could not connect to ht*p://XXX. TCP error code 10061: No connection could be made because the target machine actively refused it 192.168.XXX.XXX:XXX.
TF215097: An error occurred while initializing a build for build definition \XXX: Team Foundation services are not available from server ht*p://XXX. Technical information (for administrator): The underlying connection was closed: A connection that was expected to be kept alive was closed by the server.
TF215097: An error occurred while initializing a build for build definition \YYY: An error occurred while receiving the HTTP response to ht*p://XXX. This could be due to the service endpoint binding not using the HTTP protocol. This could also be due to an HTTP request context being aborted by the server (possibly due to the service shutting down). See server logs for more details.
Server logs provide with little info, at least we ‘ve found nothing that helps us resolve the situation. Various searches in the Net were also not productive.
Does anybody had these/similar issues? Any ideas on how/where to look for a resolution?
Thank you very much in advance for any input!
Today is a happy day, since we managed to get to the bottom of the matter. Sorry @Duat that I’m taking away the ‘answer’ checkmark – but it turned out that the problem was quite different from what you (and anybody else) has predicted.
In my last update I was about to forward the matter to MS, when we realized that our Firewall was misbehaving in the name resolution. So we assumed this was the culprit & awaited for this to resolve. After this was resolved, we STILL had the same issues and we went again re-examining the situation.
We isolated the problem within our Build Process, more specific with a custom code activity included in our build solution.
I had implemented a code activity that would kick in at the final steps of every build. This activity was about gathering
BuildDetailsabout the running build & add them as a new line in a ‘BuildLog.xls’.Implementation made use of
Microsoft.Office.Interop.Excel.This excel sheet resides in another server (NOT on the Servers where the controller/agents reside).
During development of this activity I was faced with issues like this, but after I was done no instances of EXCEL were left hanging. So I thought this was done & dealt with.
With try & error, we observed that when this activity wouldn’t ran, no problems would occur.
With this activity running, the very first build after a build-controller reset would succeed, any next build had a certain chance to fail. Once any build failed, no other would succeed until another build-controller reset.
I have only a general understanding of what the problem was (Excel-call is DCOM, TFS services are WCF : How on earth would they interfere?! Why would this sometimes succeed and sometimes fail?! ).
The provided diagnostics were no help either, in fact they mislead us into a loop that continued for months.
If I ever find the time, I ‘d like to cleanly reproduce the error & make a Server Fault question out of it…
After removal of this activity it works! I now searched in SO & found this, where J.Saunders comments: “In general, you should never use Office Interop from a server environment“.
It’s ironic that once you get to the bottom of any difficult issue, the whole universe seems to have known about it except you…