Bug 12711 - [2.5 M3 rc1][nightly-arm64][fedora 27][autobuilder2] - Event handler runqueue_stats failed
Summary: [2.5 M3 rc1][nightly-arm64][fedora 27][autobuilder2] - Event handler runqueue...
Status: RESOLVED FIXED
Alias: None
Product: Build Testing
Classification: QA/Testing
Component: general (show other bugs)
Version: 2.5
Hardware: x86 arm64
: Medium+ major
Target Milestone: 2.6 M1
Assignee: Mohamad Noor Alim
QA Contact: Aaron Chan
URL: http://centaurus02.png.intel.com:8010...
Whiteboard:
Depends on:
Blocks:
 
Reported: 2018-04-23 20:42 UTC by Aaron Chan
Modified: 2018-05-17 17:46 UTC (History)
6 users (show)

See Also:
OS type for building Yocto: ---
Type of Regression: Regression (Found by QA test)
Verified:
Documentation change: Don't know


Attachments
nightly-arm-fedora27 read-only file system (8.56 MB, application/octet-stream)
2018-04-30 02:47 UTC, Aaron Chan
no flags Details
nightly-x86_64-fedora27 read-only file system (1.29 MB, application/octet-stream)
2018-04-30 02:48 UTC, Aaron Chan
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Aaron Chan 2018-04-23 20:42:09 UTC
Build Configuration:
BB_VERSION           = "1.37.0"
BUILD_SYS            = "x86_64-linux"
NATIVELSBSTRING      = "fedora-27"
TARGET_SYS           = "aarch64-poky-linux"
MACHINE              = "qemuarm64"
DISTRO               = "poky"
DISTRO_VERSION       = "2.4+snapshot-20180423"
TUNE_FEATURES        = "aarch64"
TARGET_FPU           = ""
meta                 
meta-poky            
meta-yocto-bsp       = "master:f49ee61422c867516db252054e993df29f775136"

...
...
...

NOTE: Running task 5790 of 5843 (virtual:nativesdk:/home/ab/yocto-worker/nightly-arm64/build/meta/recipes-core/libxml/libxml2_2.9.7.bb:do_package_write_rpm)
NOTE: recipe nativesdk-opkg-1_0.3.6-r0: task do_compile: Started
NOTE: recipe nativesdk-shared-mime-info-1.9-r0: task do_package: Started
NOTE: recipe nativesdk-libxml2-2.9.7-r0: task do_package_write_rpm: Started
NOTE: recipe nativesdk-libcomps-0.1.8+gitAUTOINC+01a4759894-r0: task do_package: Started
ERROR: Execution of event handler 'runqueue_stats' failed
Traceback (most recent call last):
  File "/home/ab/yocto-worker/nightly-arm64/build/meta/classes/buildstats.bbclass", line 212, in runqueue_stats(e=<bb.event.HeartbeatEvent object at 0x7f367e536438>):
             done = isinstance(e, bb.event.BuildCompleted)
    >        system_stats.sample(e, force=done)
             if done:
  File "/home/ab/yocto-worker/nightly-arm64/build/meta/lib/buildstats.py", line 148, in SystemStats.sample(event=<bb.event.HeartbeatEvent object at 0x7f367e536438>, force=False)
OSError: [Errno 30] Read-only file system


command timed out: 16200 seconds without output running ['/home/ab/yocto-worker/nightly-arm64/yocto-autobuilder-helper/scripts/run-config', 'nightly-arm64', '/home/ab/yocto-worker/nightly-arm64/build/build', 'master', 'git://git.yoctoproject.org/poky', '', 'None', 'None', 'http://centaurus02.png.intel.com:8010/#builders/19/builds/12'], attempting to kill
process killed by signal 9
program finished with exit code -1


After performing the QA test on nightly-arm64, the Fedora27 KVM distribution file-system becomes read-only and once the KVM has been re instantiated, it went into emergency mode
and unable to boot Kernel.
Comment 1 Paul Eggleton 2018-04-24 03:51:52 UTC
So unless I'm missing something this is an infrastructure issue - right? The file it's writing to is under BUILDSTATS_BASE which is under TMPDIR and thus shouldn't ever be read-only for the build system.
Comment 2 Richard Purdie 2018-04-26 07:41:54 UTC
See Paul's question
Comment 3 Aaron Chan 2018-04-26 18:53:43 UTC
Hi Paul,

We are trying to resolve this as the entire KVM Fedora27 file-system became read-only and system ran into emergency mode after we ran oe-selftest. I am trying to get Alim to recommission the entire Distro and rerun again.
Comment 4 Paul Eggleton 2018-04-29 13:18:49 UTC
What could we be doing in the tests though that would cause the system to go into read-only mode?
Comment 5 Aaron Chan 2018-04-30 02:45:53 UTC
There are 2 occurrence which results in read-only file system during bitbake on fedora27.

nightly-arm64 (first occurance)
-------------
Link: http://centaurus02.png.intel.com:8010/#/builders/19/builds/12

NOTE: recipe nativesdk-libcomps-0.1.8+gitAUTOINC+01a4759894-r0: task do_package: Started
ERROR: Execution of event handler 'runqueue_stats' failed
Traceback (most recent call last):
  File "/home/ab/yocto-worker/nightly-arm64/build/meta/classes/buildstats.bbclass", line 212, in runqueue_stats(e=<bb.event.HeartbeatEvent object at 0x7f367e536438>):
             done = isinstance(e, bb.event.BuildCompleted)
    >        system_stats.sample(e, force=done)
             if done:
  File "/home/ab/yocto-worker/nightly-arm64/build/meta/lib/buildstats.py", line 148, in SystemStats.sample(event=<bb.event.HeartbeatEvent object at 0x7f367e536438>, force=False)
OSError: [Errno 30] Read-only file system

nightly-x86_64 (2nd occurance)
--------------
Link: http://centaurus02.png.intel.com:8010/#/builders/24/builds/4

NOTE: recipe core-image-sato-1.0-r0: task do_image_wic: Started
ERROR: Execution of event handler 'runqueue_stats' failed
Traceback (most recent call last):
  File "/home/ab/yocto-worker/nightly-x86-64/build/meta/classes/buildstats.bbclass", line 212, in runqueue_stats(e=<bb.event.HeartbeatEvent object at 0x7fc4211c61d0>):
             done = isinstance(e, bb.event.BuildCompleted)
    >        system_stats.sample(e, force=done)
             if done:
  File "/home/ab/yocto-worker/nightly-x86-64/build/meta/lib/buildstats.py", line 148, in SystemStats.sample(event=<bb.event.HeartbeatEvent object at 0x7fc4211c61d0>, force=False):
                                      data +
    >                                 b'\n')
                 self.last_proc = now
OSError: [Errno 30] Read-only file system

We have completed recommission the entire KVM Distro Fedora27 assuming this is an Infrastructure issue, 
it shouldn't reoccur again as we seen such abnormalities in other distribution QA cycle but was never recorded anywhere. 
This ticket is to track this kind of abnormalities which may or may not be caused by bitbake or whatever.

We ran a sanity check with running Oe-selftest excluding Bitbake and it PASSED successfully.

Link: http://centaurus02.png.intel.com:8010/#/builders/18/builds/25
Comment 6 Aaron Chan 2018-04-30 02:47:36 UTC
Created attachment 4292 [details]
nightly-arm-fedora27 read-only file system
Comment 7 Aaron Chan 2018-04-30 02:48:47 UTC
Created attachment 4293 [details]
nightly-x86_64-fedora27 read-only file system
Comment 8 Mohamad Noor Alim 2018-05-17 17:46:32 UTC
I fixed this issue by recomission new KVM for Fedora 27. I also tested on 2.5 M4 rc1 and didn't see any issue regarding to read-only file system.