Project

General

Profile

Actions

Bug #12936

closed

shec fails i386 make check

Added by Sage Weil over 8 years ago. Updated over 8 years ago.

Status:
Resolved
Priority:
Urgent
Assignee:
Category:
-
Target version:
-
% Done:

0%

Source:
Q/A
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

http://gitbuilder.sepia.ceph.com/gitbuilder-ceph-tarball-trusty-i386-basic/log.cgi?log=c938d1f95e79ab5541c453b070fd97e44623781c

../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:33: non_regression ceph_erasure_code_non_regression --check --stripe-width 4096 --plugin shec --parameter technique=multiple --parameter k=6 --parameter m=3 --parameter c=2 true --base ../ceph-erasure-code-corpus/v0.92-988
chunk 6 encodes differently
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:33: non_regression return 1


Related issues 1 (0 open1 closed)

Related to Ceph - Bug #12933: erasure-code: non regression fails to load pluginResolvedLoïc Dachary09/03/2015

Actions
Actions #1

Updated by Loïc Dachary over 8 years ago

  • Status changed from New to Resolved
Actions #2

Updated by Sage Weil over 8 years ago

  • Status changed from Resolved to 12

This still happens:

../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:84: test_shec for stripe_width in '$STRIPE_WIDTHS'
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:85: test_shec shec_action --check --stripe-width 4096 --plugin shec --parameter technique=multiple --parameter k=6 --parameter m=3 --parameter c=2 true --base ../ceph-erasure-code-corpus/v0.92-988
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:64: shec_action local action=--check
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:65: shec_action shift
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:67: shec_action non_regression --check --stripe-width 4096 --plugin shec --parameter technique=multiple --parameter k=6 --parameter m=3 --parameter c=2 true --base ../ceph-erasure-code-corpus/v0.92-988
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:29: non_regression local action=--check
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:30: non_regression shift
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:32: non_regression test --check '!=' NOOP
../ceph-erasure-code-corpus/v0.92-988/non-regression.sh:33: non_regression ceph_erasure_code_non_regression --check --stripe-width 4096 --plugin shec --parameter technique=multiple --parameter k=6 --parameter m=3 --parameter c=2 true --base ../ceph-erasure-code-corpus/v0.92-988
chunk 6 encodes differently

at http://gitbuilder.sepia.ceph.com/gitbuilder-ceph-tarball-trusty-i386-basic/log.cgi?log=27d1e4d41e0a345dce96ecab620fefb1c900acca

Actions #3

Updated by Loïc Dachary over 8 years ago

  • Assignee set to Loïc Dachary
Actions #4

Updated by Loïc Dachary over 8 years ago

  • Status changed from 12 to In Progress
Actions #5

Updated by Loïc Dachary over 8 years ago

  • Status changed from In Progress to 12
Actions #6

Updated by Greg Farnum over 8 years ago

If this is going to take some time can we back out the test (...or the plugin? :/) so that any new errors in i386 are immediately visible? Red lines are bad for our quality and detection abilities.

Actions #7

Updated by Greg Farnum over 8 years ago

And just in case anybody else tries searching for it with the same string I did:, this is failing the qa/workunits/erasure-code/encode-decode-non-regression.sh script.

Actions #8

Updated by Loïc Dachary over 8 years ago

  • Status changed from 12 to In Progress
Actions #9

Updated by Loïc Dachary over 8 years ago

Fail for 6/3/2 and 9/4/2 only

Actions #10

Updated by Loïc Dachary over 8 years ago

  • valgrind is happy
  • the bug disapear when adding print statements in some places
  • the bug disapear when the code is not optimized and shows up when the code is -O2 or -O3
  • the bug stays when running gdb but introspection is not possible because all variables are optimized out
Actions #11

Updated by Loïc Dachary over 8 years ago

With non optimized code on i386 (or on amd64):

ceph_erasure_code_non_regression --check --debug-osd=20 --stripe-width 4096 --plugin shec --para\
meter technique=multiple --parameter k=6 --parameter m=3 --parameter c=2 true --base ../ceph-erasure-code-corpus/v0.92-988
2015-09-21 13:26:34.176415 b606f700 10 ErasureCodePluginSelectShec: load: shec_generic
2015-09-21 13:26:34.176438 b606f700 10 ErasureCodePluginSelectShec: generic plugin
2015-09-21 13:26:34.176472 b606f700 10 ErasureCodeShec: (k, m, c) set to (6, 3, 2)
2015-09-21 13:26:34.176483 b606f700 10 ErasureCodeShec: w default to 8
2015-09-21 13:26:34.176501 b606f700 10 ErasureCodeShec: [ cache tables ] creating coeff for k=6 m=3 c=2 w=8
2015-09-21 13:26:34.176542 b606f700 10 ErasureCodeShec: matrix =
2015-09-21 13:26:34.176543 b606f700 10 ErasureCodeShec: 111111
2015-09-21 13:26:34.176544 b606f700 10 ErasureCodeShec: 111000
2015-09-21 13:26:34.176545 b606f700 10 ErasureCodeShec: 000111
2015-09-21 13:26:34.176545 b606f700 10 ErasureCodeShec:  [ technique ] = multiple
2015-09-21 13:26:34.176547 b606f700 10 ErasureCodePluginShec: ErasureCodePluginShec: factory() completed
2015-09-21 13:26:34.176846 b606f700 20 ErasureCodeShecTableCache: [ get table    ] = 17600744530118
2015-09-21 13:26:34.176873 b606f700 20 ErasureCodeShecTableCache: [ put table    ] = 17600744530118
2015-09-21 13:26:34.176875 b606f700 20 ErasureCodeShecTableCache: [ store table  ] = 17600744530118
2015-09-21 13:26:34.176878 b606f700 20 ErasureCodeShecTableCache: [ cache size   ] = 1
2015-09-21 13:26:34.176894 b606f700 20 ErasureCodeShecTableCache: [ get table    ] = 5506358495420614
2015-09-21 13:26:34.176898 b606f700 20 ErasureCodeShecTableCache: [ put table    ] = 5506358495420614
2015-09-21 13:26:34.176899 b606f700 20 ErasureCodeShecTableCache: [ store table  ] = 5506358495420614
2015-09-21 13:26:34.176901 b606f700 20 ErasureCodeShecTableCache: [ cache size   ] = 2

With optimized (-O2) code on i386
2015-09-21 13:23:28.402536 b605a700 10 ErasureCodePluginSelectShec: load: shec_generic
2015-09-21 13:23:28.402564 b605a700 10 ErasureCodePluginSelectShec: generic plugin
2015-09-21 13:23:28.402595 b605a700 10 ErasureCodeShec: (k, m, c) set to (6, 3, 2)
2015-09-21 13:23:28.402610 b605a700 10 ErasureCodeShec: w default to 8
2015-09-21 13:23:28.402627 b605a700 10 chunk 6 encodes differently
ErasureCodeShec: [ cache tables ] creating coeff for k=6 m=3 c=2 w=8
2015-09-21 13:23:28.402665 b605a700 10 ErasureCodeShec: matrix =
2015-09-21 13:23:28.402666 b605a700 10 ErasureCodeShec: 111000
2015-09-21 13:23:28.402666 b605a700 10 ErasureCodeShec: 000111
2015-09-21 13:23:28.402667 b605a700 10 ErasureCodeShec: 111111
2015-09-21 13:23:28.402668 b605a700 10 ErasureCodeShec:  [ technique ] = multiple
2015-09-21 13:23:28.402670 b605a700 10 ErasureCodePluginShec: ErasureCodePluginShec: factory() completed

Actions #12

Updated by Loïc Dachary over 8 years ago

  • Status changed from In Progress to Fix Under Review
Actions #13

Updated by Loïc Dachary over 8 years ago

  • Status changed from Fix Under Review to In Progress
Actions #14

Updated by Sage Weil over 8 years ago

  • Priority changed from Urgent to High
Actions #15

Updated by Loïc Dachary over 8 years ago

  • Status changed from In Progress to Fix Under Review
  • Priority changed from High to Urgent
Actions #16

Updated by Sage Weil over 8 years ago

  • Status changed from Fix Under Review to Resolved
Actions

Also available in: Atom PDF