intermittent failure github.com/juju/juju/featuretests cmd_juju_user_test.go:82: UserSuite.TestUserResetPasswordForOther

Bug #1763328 reported by John A Meinel on 2018-04-12
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
juju
Medium
Unassigned

Bug Description

As seen here:
  http://10.125.0.203:8080/job/RunUnittests-arm64/357/testReport/junit/github/com_juju_juju_featuretests/TestPackage/
cmd_juju_user_test.go:90:
    c.Assert(cmdtesting.Stderr(context), gc.Matches, `
    Password for "bob" has been reset.
    Ask the user to run:
         juju register (.+)
    `[1:])
... value string = "" +
... "WARNING http: TLS handshake error from 127.0.0.1:36638: EOF\n" +
... "\n" +
... "Password for \"bob\" has been reset.\n" +
... "Ask the user to run:\n" +
... " juju register MEQTA2JvYjAREw9sb2NhbGhvc3Q6NDU5ODMEIFqmrGsMCeoOr3CSMutNueyxxiw9kS-pt8wHnyvxsZDPEwhrb250cm9sbAAA\n"
... regex string = "" +
... "Password for \"bob\" has been reset.\n" +
... "Ask the user to run:\n" +
... " juju register (.+)\n"

The issue seems to be that we got a TLS handshake error before connecting, and that led to a WARNING being issued on stderr/stdout before the expected output.

Oddly, it seems to be the last thing in the log, but the first in the output:
[LOG] 0:01.127 INFO juju.cmd.modelcmd setting controllerName to "" true
[LOG] 0:01.300 WARNING juju.apiserver http: TLS handshake error from 127.0.0.1:36638: EOF

John A Meinel (jameinel) wrote :

It may be a buffering issue, as far as where the WARNING happens (if it happens on stderr, which isn't buffered, while stdout is buffered).

We *could* change the test so it only reads stdout, but arguably something is really going wrong if we're getting a TLS handshake failure. Are we connecting a second time and we've already started tearing down the server?

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers