Project

General

Profile

action #81142

VNC console corruption

Added by MDoucha 10 months ago. Updated 3 days ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
Concrete Bugs
Target version:
Start date:
2020-12-17
Due date:
% Done:

0%

Estimated time:
Difficulty:

Description

It appears that issue #61994 is back, this time on PPC64LE. LTP tests fail to boot because switching from boot animation to login screen corrupts VNC frame buffer and needles fail to match.
https://openqa.suse.de/tests/5187741#step/boot_ltp/9
https://openqa.suse.de/tests/5187758#step/boot_ltp/8

framebuffer-6794466.png (28.3 KB) framebuffer-6794466.png MDoucha, 2021-08-11 15:55
11771

Related issues

Related to openQA Tests - action #81808: [sle][security][sle15sp3] test fails in ima_appraisal_hashes about background screen chaosResolved2021-01-06

Related to openQA Infrastructure - action #61994: VNC console corruption on aarch64Resolved2020-01-102020-01-21

History

#1 Updated by pvorel 10 months ago

  • Subject changed from VNC console corruption on PPC64LE to VNC console corruption

#2 Updated by rfan1 10 months ago

#3 Updated by pvorel 10 months ago

  • Related to action #81808: [sle][security][sle15sp3] test fails in ima_appraisal_hashes about background screen chaos added

#4 Updated by coolo 10 months ago

  • Project changed from openQA Infrastructure to openQA Project

I don't think this is an infrastructure problem - until proven differently. Do we have any test case that shows it often enough to justify any debugging on this?

#5 Updated by MDoucha 10 months ago

coolo wrote:

I don't think this is an infrastructure problem - until proven differently. Do we have any test case that shows it often enough to justify any debugging on this?

We have no specific test case. But we do have multiple random test failures all over OSD every day now due to this issue.

But I guess what might trigger this bug reliably would be booting to desktop and then sending alternately Ctrl+Alt+F1 and Alt+F7 with 5 second delay about a hundred times.

#6 Updated by okurz 9 months ago

  • Target version changed from Ready to future

I don't think SUSE QE Tools can help easily here. I am convinved this needs better reproduction from test side. If you can achieve that and need help in something backend related where SUSE QE Tools can likely help we will look into this again.

#7 Updated by okurz 9 months ago

  • Category set to Concrete Bugs

#8 Updated by vpelcak 9 months ago

Hello.

Martin Doucha will start working on reproducer for you towards the end of the week.

#9 Updated by szarate 8 months ago

#10 Updated by MDoucha 8 months ago

szarate wrote:

I wonder if it's related to: https://bugzilla.opensuse.org/show_bug.cgi?id=1160707

Yes, it appears to be the same bug. Also reported as https://progress.opensuse.org/issues/61994 the first time it appeared a year ago on aarch64.

#11 Updated by MDoucha 8 months ago

  • Related to action #61994: VNC console corruption on aarch64 added

#12 Updated by MDoucha 8 months ago

I've tried to reliably reproduce the issue by writing a test module that'll just switch between console login screen and X11 login screen and check needles after each switch but I've failed so far. When was the last time the VNC corruption happened?

#13 Updated by acarvajal 3 months ago

Seen this (or something similar) three times today in SLES+HA QAM Single Incidents 15-SP3 job group:

Issue seems sporadic, as test is passing with other recent MUs. Last jobs even has 3 fence/reboots and issue is present only the third time.

#14 Updated by MDoucha 3 months ago

I've created PR https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/13020 to rule out one possible cause of the console corruption. Let's see what the kernel thinks the console should look like.

#15 Updated by MDoucha 2 months ago

11771

We have first framebuffer screenshots of the screen corruption bug (PNG version attached):
https://openqa.suse.de/tests/6794466#step/boot_ltp/4

Result: negative.

Which means that the bug is somewhere between bochsdrmfb driver and VNC client. Next, I'll try switching QEMUVGA device in kernel tests to see whether the bug shows up anyway by September.

#16 Updated by openqa_review about 2 months ago

This is an autogenerated message for openQA integration by the openqa_review script:

This bug is still referenced in a failing openQA test: qam_ha_priority_fencing_node02
https://openqa.suse.de/tests/6639074

To prevent further reminder comments one of the following options should be followed:

  1. The test scenario is fixed by applying the bug fix to the tested product or the test is adjusted
  2. The openQA job group is moved to "Released" or "EOL" (End-of-Life)
  3. The label in the openQA scenario is removed

#17 Updated by MDoucha 3 days ago

I've reported kernel/QEMU bug after further investigation:
https://bugzilla.suse.com/show_bug.cgi?id=1191812

Also available in: Atom PDF