lava_test_shell connection dropped on highbank and midway when performing network test

Bug #1274491 reported by Milosz Wasilewski
14
This bug affects 1 person
Affects Status Importance Assigned to Milestone
LAVA Dispatcher
Fix Released
High
Dave Pigott

Bug Description

https://validation.linaro.org/scheduler/job/107375/log_file#L_74_0
Lava hangs on performing network-test-basic on highbank. Subsequent tests are skipped. The test sequence can be repeated manually during hacking session - no hang observed. The problem started on Jan 24th (highbank build 594)

Revision history for this message
Botao (botao-sun) wrote :

Please refer to attachment to find the log for manually run in LAVA hacking session.

Changed in lava-dispatcher:
status: New → Confirmed
importance: Undecided → High
assignee: nobody → Dave Pigott (dpigott)
Revision history for this message
Botao (botao-sun) wrote :

Same issue is observed on "midway" test result started from build 600:

https://validation.linaro.org/dashboard/image-reports/linux-linaro-midway

https://validation.linaro.org/scheduler/job/108292/log_file#L_78_0

<LAVA_DISPATCHER>2014-02-01 03:37:26 AM WARNING: lava_test_shell connection dropped

Full log can be found here:

https://validation.linaro.org/scheduler/job/108292/log_file

tags: added: highbank linaro-ubuntu linux-linaro midway qa-services
summary: - lava_test_shell connection dropped on highbank when performing network
- test
+ lava_test_shell connection dropped on highbank and midway when
+ performing network test
Revision history for this message
Tyler Baker (tyler-baker) wrote :

Hi Dave,

Any progress made on this bug?

Revision history for this message
Dave Pigott (dpigott) wrote :

Haven't had a chance to look, but this is a known problem with Calxeda nodes. They get into a weird power state and can't be controlled. It usually takes a while and then the node starts responding. In the past I've fixed it with a power status query, so maybe I'll look at adding that into the high bank dispatcher code and trying it out.

Revision history for this message
Dave Pigott (dpigott) wrote :
Changed in lava-dispatcher:
status: Confirmed → In Progress
Revision history for this message
Dave Pigott (dpigott) wrote :

A slight mistake in my original fix - now fixed and tested in staging

Changed in lava-dispatcher:
status: In Progress → Fix Committed
Changed in lava-dispatcher:
status: Fix Committed → Fix Released
Revision history for this message
Botao (botao-sun) wrote :

Hi Dave and Tyler, looks like change 1186 hasn't fixed this issue:

https://validation.linaro.org/dashboard/image-reports/linux-linaro-highbank

https://validation.linaro.org/dashboard/image-reports/linux-linaro-midway

Test still be dropped on 23 March and 25 March, but that change had been merged on 21 March. Has it been published to the production environment? I remember there was a case that the change could only be deployed in a LAVA upgrade event.

Revision history for this message
Botao (botao-sun) wrote :

@Tyler, @Dave, would you please change the status of this bug to "Fix Committed"? I ask because this issue still exists on the latest test result in LAVA, for both Highbank and Midway (#626):

https://validation.linaro.org/dashboard/image-reports/linux-linaro-highbank

https://validation.linaro.org/dashboard/image-reports/linux-linaro-midway

So it actually should not be considered as "Fix Released", thanks.

Revision history for this message
Botao (botao-sun) wrote :

For the test in LAVA happened on 24 April 2014 with the build #630 (same as the build ran on 21 April 2014), the issue has gone:

https://validation.linaro.org/dashboard/streams/private/team/linaro/pre-built-midway/bundles/78303c66ecf2382bf04375505ac7f407042f841b/

But let's wait for 1 more build to see what will happen.

Revision history for this message
Botao (botao-sun) wrote :

Now Highbank and Midway have been removed from Linux Linaro ubuntu daily test services.

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

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.