Hi all,

On Monday I was doing some manual testing on Cindy's GPII-2662 pull request 
(https://github.com/GPII/universal/pull/582 -- now merged) and wanted to test 
the universal changes against the local flow manager running on Windows.

I grabbed the latest https://github.com/GPII/windows master and modified the 
package.json to point to Cindy's branch:

$ git diff
diff --git a/package.json b/package.json
index 7184384..e920c9d 100644
--- a/package.json
+++ b/package.json
@@ -12,7 +12,7 @@
         "ref-array": "1.1.2",
         "edge": "6.5.1",
         "string-argv": "0.0.2",
-        "universal": "GPII/universal#694422fc218ed6fdb31278242e67cf2fec8b8901"
+        "universal": "cindyli/universal#GPII-2662"
     },
     "devDependencies": {
         "grunt": "1.0.1",

And ran vagrant up.

The provisioning process did not complete and appeared to get stuck. There was 
no error from the vagrant process, just that there was no further advance. 
Unfortunately, I didn't record the output or time that run. But I estimate that 
I terminated vagrant after 30-45 mins.

I then ran vagrant up 3 more times and did time and record the output:

0. The run described above: windows project with Cindy's universal
        - Stuck, terminated after 30-45 mins
1. The windows project as is from github without any of Cindy's work
        - https://gist.github.com/simonbates/694a2017aefc4e30501686d07cb9e4e0
        - Success, total time: 25 mins
2. Same run as the initial one, with Cindy's universal work
        - https://gist.github.com/simonbates/272daa3a14a24372bbdd5f14f3f4b8e8
        - Stuck, terminated after 50 mins
        - I think that it got stuck at a point different from run 0 (but I 
don't have that run's output to confirm)
3. Same run again, with Cindy's universal work
        - https://gist.github.com/simonbates/97a03747de8b99029240c8090a832477
        - Success, total time: 27 mins

Observations:

- The provisioning takes a long time
- The provisioning, in my experience with a small sample size, is 50% successful
        - It's difficult to tell if using Cindy's universal (together with all 
the other changes that have happened on universal since the one that the 
windows repo uses) had any impact (would need more runs to determine)
- There is no indication of why the provisioning failed

These issues together make using the windows Vagrantfile quite difficult.

I spoke briefly to Joseph and it appears that he has seen similar reliability 
issues and I'm wondering if others have also?

Is it possible to speed up the provisioning?

Do we have any ideas why it might be unreliable?

I was going to raise this at the architecture meeting today but I won't be able 
to make it due to another appointment.

Machine details:

- Linux host, Ubuntu 16.04
- Vagrant 2.0.1
- VirtualBox 5.1.32
- latest Windows image

Simon

_______________________________________________
Architecture mailing list
[email protected]
https://lists.gpii.net/mailman/listinfo/architecture

Reply via email to