Bug 1367657

Created attachment 1191473 [details]
invalid_window_blocks_registration_to_ppc64_in_GUI.png

Created attachment 1191474 [details]
rhsm.log

The bug happened on RHEL-7.3-server-ppc64le-Beta-1.1 when register the system second time in GUI against Satellite-6.2.0-RHEL-7-20160817.0

Please attach the foreman-debug output from the satellite server when this occurred.

I tried again both on RHEL-7.3-server-ppc64le-Beta-1.1 and RHEL-7.3-server-ppc64-Beta-1.1, but this bug can not be reproduced. So close this bug As WORKSFORME. If this bug happens again, I will attach the foreman-debug output.

This bug is reproduced with RHEL-7.3-server-ppc64-snapshot-2.0 and Satellite-6.2.0-RHEL-7-20160831.0, so reopen it. 
I ran 'foreman-debug' on satellite after this bug happened again. Please see attachment 'foreman-debug-U57YK.tar.xz'.

Created attachment 1198066 [details]
foreman-debug-U57YK.tar.xz

We see there is an issue. Can you please run `rhsm-debug system` and attach the resulting zip file? Also, if we could get access to the system in order to reproduce/validate/debug, that would be helpful.

Note for dev: fix the way the GUI responds to the 500, the 500 error can be fixed with bug 1367676. Additionally, initial high priority is to validate that the error box is not a regression in 7.3.

I can reproduce this by breaking my registergui.py. Specifically, inside the try block of _register_consumer of the registergui.py I place the following:
"""
raise RestlibException(404, "Problem creating unit Consumer<id=8a90c4a8568da2550156972d1aaf03bd, uuid=8e6aadbf-6d57-40b0-8dff-3e2822627425, type=system, name=ibm-p8-kvm-05-guest-01.rhts.eng.bos.redhat.com>")
"""

The issue appears to be with the way we are placing the message contained within RestlibExceptions into a GTK parsed string that already contains markup.

The following is printed to the cli when the empty dialog box appears:

"""
(subscription-manager-gui:7055): Gtk-WARNING **: Failed to set text from markup due to error parsing markup: Error on line 2 char 35: Odd character '=', expected a '>' or '/' character to end the start tag of element 'id', or optionally an attribute; perhaps you used an invalid character in an attribute name
"""


A potential fix would be to sanitize error messages received from candlepin. (Remove '<>' and friends).

(In reply to Kevin Howell from comment #9)
> We see there is an issue. Can you please run `rhsm-debug system` and attach
> the resulting zip file? Also, if we could get access to the system in order
> to reproduce/validate/debug, that would be helpful.

This bug is not always reproduced. I tried 2 times to reproduce it, but failed today. I prefer to keep it as NEW and retest it with next RHEL7.3 build.

I have a PR up that should fix this issue (linked by the github tracker). It is not always reproducible because reproducing it depends entirely on candlepin failing in the right way to give an error message to be displayed by subman that includes characters that look like markup (as above).

I tried RHEL-7.3-server-ppc64-20160908.3 and RHEL-7.3-server-ppc64le-20160908.3 against Satellite-6.2.0-RHEL-7-20160831.0, and failed to reproduce this bug:

[root@ibm-p730-03-lp3 ~]# subscription-manager version
server type: Red Hat Subscription Management
subscription management server: 3.0.0.78-Unknown
subscription management rules: 5.15.2
subscription-manager: 1.17.14-1.el7
python-rhsm: 1.17.9-1.el7

So move it from ON_QA to VERIFIED. If this bug is reproduced again, I will reopen it.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHSA-2016-2592.html