Hi
I’m using the latest version 1.0.0.1557 on a Windows 2008 server.
1. start manual build of code from continua and builds as expected.
2. push new changes to GIT
3. verify changes are in correct branch using separate GIT client
4. start manual build, this builds with the same version as in step 1??
5. Go into repository config and click reset repository
6. start build manually, builds correct (latest) version of code
I have tried waiting for over 60 mins after pushing code but still the old code is built up to the point of me running a Repository reset. Is there a caching setting? or is this a bug?
Thanks
Rob
Apologies, this should be in the Continua Forum 
Hi Rob
This sounds like a bug. Can you tell us what settings you have on the repository so we can try to duplicate it here?
Hi Rob
I’m not able to reproduce this here, can you confirm what settings you have?
Sorry I have not been back sooner… I have just ran again, and experienced the same issue, so here is some further information.
v1.0.0.1557
Project Repository - Type = GIT
Name = UAT-LIVE
Polling = 60 Secs
Timeout = 60
Type = Git
Repository = http://our.externalurl.com/scm/project.git
Username = AD_UserServiceAccount
Password = *************
Branches to Monitor = Single Branch
Branch Name = v5.2.0
Recursive = (un checked)
Using = Git.Default
Enabled = Checked
Validate = OK
Exclude Patterns = none set
Downtime = none set
So again to fix I went to Admin / CI Server / Repositories / Reset (wait about 5 mins to complete)
Then built from latest check-in ok.
I see there have been 2 more releases since 1.0.0.1557, so I’m going to upgrade to the latest and re-test
Hi Rob
If you see this again, check the eventlog in Continua CI and also the windows eventlog, and also in the config wizard, check if the repository is showing as errored. Is it possible your access (ie internet/network) to the repository server goes down at some point every day (for backups perhaps?). If so, then configuring downtime on the Continua repository may resolve this. We’ll do some more testing with git again today.
Hi Vincent
Thanks for your reply. Since upgrading to 1584 we still seem to be having an issue with this. The issue happens everytime at different times of the day. Once we reset the repository then it works okay. Therefor I do not believe this is down to a backup or being offline.
I have tried to enable logging but with limited success, by uncommenting the system diagnostics section in the file Continua.Server.Service.exe.config log I get a trace file. But I struggle to understand / find relevent information in this. I have increased the Windows Event Service to Debug in the logger section and restarted the windows service but the Windows Event Log only reports the service stopping and started. I have looked in the Event Log within continua, but again this information is limited. Perhaps I need to enable something else in the configs?
I have attached the trace file for your information.
NOTE it may also show useful information regarding another issue I have logged this morning regarding passing variables back from batch files.
Hi
Just to give you an update, I have downloaded and installed 1.0.0.1603, then rebooted the server. Unfortunately I still have to reset the repository to for the configuration to pickup the latest check-ins (git commit / git push).
Rob
Me again…
On my latest ‘reset’ the repository reported an ‘error whilst initializing’ which it has not done before. In the continua event log I see
Cannot get new changesets for @link(2002, d4e4d5c4-4c53-4729-a04e-e5d1b458c4dd)[UAT-LIVE]: Running git.exe failed with return code: 128 and error output: error: The requested URL returned error: 500 while accessing http://xxxxxxxxxxxxxx:********************@stash.xxxxxxxxx.xxx/scm/xxxxxxxxxx.git/info/refs?service=git-upload-pack
fatal: HTTP request failed
I assume Continua received a http 500 error, so I’m going to test and investigate this.
Rob
Hi Rob
I have just managed to reproduce the problem with not detecting changes, running a new build right now with the fix.
Just to confirm, this is now resolved. Tx Rob
Thanks Rob.