Project

General

Profile

Bug #40018

crash in io_context thread when lots of connections abort

Added by Ali Maredia over 3 years ago. Updated about 2 years ago.

Status:
Resolved
Priority:
High
Assignee:
Abhishek Lekshmanan
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
luminous, mimic, nautilus
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
29967
Crash signature (v1):
Crash signature (v2):

Description

During a small percentage of runs of the full rgw:verify suite the radosgw crashes in the middle of a run of the java s3-tests suite.

See .txt files with snippits of teuthology logs and the link to the logs themselves.

java_s3_tests_crash_log_1.txt View (238 KB) Ali Maredia, 05/23/2019 09:47 PM

java_s3_tests_crash_log_2.txt View (100 KB) Ali Maredia, 05/23/2019 09:47 PM

Related issues

Related to rgw - Bug #57784: beast frontend crashes on exception from socket.local_endpoint() Pending Backport
Copied to rgw - Backport #41569: nautilus: crash in io_context thread when lots of connections abort Resolved
Copied to rgw - Backport #41570: mimic: crash in io_context thread when lots of connections abort Resolved
Copied to rgw - Backport #41571: luminous: crash in io_context thread when lots of connections abort Rejected

History

#2 Updated by Matt Benjamin over 3 years ago

  • Status changed from New to Triaged
  • Assignee set to Casey Bodley

#3 Updated by Abhishek Lekshmanan over 3 years ago

  • Priority changed from Normal to High

#4 Updated by Abhishek Lekshmanan over 3 years ago

it looks like the io_context::run() ec overload variant is deprecated. https://www.boost.org/doc/libs/1_66_0/doc/html/boost_asio/reference/io_context/run/overload2.html

Following down the stacktrace, it seems like the ec value is ignored in boost::asio::detail::scheduler_operation::complete

#5 Updated by Casey Bodley over 3 years ago

  • Subject changed from RGW is crashing after some Java S3 tests to crash in io_context thread when lots of connections abort

#6 Updated by Abhishek Lekshmanan over 3 years ago

  • Status changed from Triaged to In Progress
  • Assignee changed from Casey Bodley to Abhishek Lekshmanan

#7 Updated by Nathan Cutler over 3 years ago

  • Status changed from In Progress to Pending Backport
  • Backport set to mimic, nautilus
  • Pull request ID set to 29967

#8 Updated by Nathan Cutler over 3 years ago

  • Copied to Backport #41569: nautilus: crash in io_context thread when lots of connections abort added

#9 Updated by Nathan Cutler over 3 years ago

  • Copied to Backport #41570: mimic: crash in io_context thread when lots of connections abort added

#10 Updated by Nathan Cutler over 3 years ago

  • Backport changed from mimic, nautilus to luminous, mimic, nautilus

#11 Updated by Nathan Cutler over 3 years ago

  • Copied to Backport #41571: luminous: crash in io_context thread when lots of connections abort added

#12 Updated by Nathan Cutler about 2 years ago

  • Status changed from Pending Backport to Resolved

While running with --resolve-parent, the script "backport-create-issue" noticed that all backports of this issue are in status "Resolved" or "Rejected".

#13 Updated by Casey Bodley 4 months ago

  • Related to Bug #57784: beast frontend crashes on exception from socket.local_endpoint() added

Also available in: Atom PDF