[MESOS-1892] Using mesos-0.20.1.jar with libmesos-0.21.0 reliably segfaults

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Cannot Reproduce
• Affects Version/s: 0.21.0
• Fix Version/s: 0.21.0
• Component/s: java api

Environment

Observed on Ubuntu 14.04.1 LTS (trusty) using both OpenJDK6 and OpenJDK7.

• Target Version/s: 0.21.0
• Assignee: Timothy Chen
• Reporter: Connor Doyle
• Created: 09/Oct/14 19:05
• Updated: 23/Oct/14 00:39
• Resolved: 23/Oct/14 00:39

Description

The segfault seems to occur on every call to driver.launchTasks()

Here is a minimal(ish) Java framework for testing purposes:
https://github.com/ConnorDoyle/mesos-compat

Here is some output from running the framework against a Mesos 0.20.1 cluster with libmesos-0.20.1:

mesosMaster: [127.0.1.1:5050]
I1009 18:22:28.044186 20570 sched.cpp:139] Version: 0.20.1
I1009 18:22:28.045871 20564 sched.cpp:235] New master detected at master@127.0.1.1:5050
I1009 18:22:28.048756 20564 sched.cpp:243] No credentials provided. Attempting to register without authentication
I1009 18:22:28.050904 20564 sched.cpp:409] Framework registered with 20141009-161859-16842879-5050-13592-0006
Registered with the Mesos master:
id: "20141009-161859-16842879-5050-13592"
ip: 16842879
port: 5050
pid: "master@127.0.1.1:5050"
hostname: "mesos.vm"

Assigned framework ID:
value: "20141009-161859-16842879-5050-13592-0006"

Received resource offers:
id {
  value: "20141009-161859-16842879-5050-13592-9"
}
framework_id {
  value: "20141009-161859-16842879-5050-13592-0006"
}
slave_id {
  value: "20141006-070645-16842879-5050-1540-0"
}
hostname: "10.141.141.10"
resources {
  name: "cpus"
  type: SCALAR
  scalar {
    value: 1.0
  }
  role: "*"
}
resources {
  name: "mem"
  type: SCALAR
  scalar {
    value: 1001.0
  }
  role: "*"
}
resources {
  name: "disk"
  type: SCALAR
  scalar {
    value: 34068.0
  }
  role: "*"
}
resources {
  name: "ports"
  type: RANGES
  ranges {
    range {
      begin: 31000
      end: 32000
    }
  }
  role: "*"
}

Building task list...
Calling SchedulerDriver.launchTasks...
Received a status update:
task_id {
  value: "mesos-compat-0"
}
state: TASK_RUNNING
slave_id {
  value: "20141006-070645-16842879-5050-1540-0"
}
timestamp: 1.412878948485535E9

Received a status update:
task_id {
  value: "mesos-compat-0"
}
state: TASK_FINISHED
message: "Command exited with status 0"
slave_id {
  value: "20141006-070645-16842879-5050-1540-0"
}
timestamp: 1.412878949486471E9

Here is some output from running the framework against a Mesos 0.21.0-SNAPSHOT cluster with libmesos-0.21.0-SNAPSHOT:

mesosMaster: [172.31.38.219:5050]
I1009 18:34:01.391448 21087 sched.cpp:137] Version: 0.21.0
I1009 18:34:01.394507 21081 sched.cpp:233] New master detected at master@172.31.38.219:5050
I1009 18:34:01.394839 21081 sched.cpp:241] No credentials provided. Attempting to register without authentication
I1009 18:34:01.403170 21084 sched.cpp:407] Framework registered with 20141008-182754-3676708780-5050-27608-0002
Registered with the Mesos master:
id: "20141008-182754-3676708780-5050-27608"
ip: 3676708780
port: 5050
pid: "master@172.31.38.219:5050"
hostname: "ip-172-31-38-219.eu-west-1.compute.internal"

Assigned framework ID:
value: "20141008-182754-3676708780-5050-27608-0002"

Received resource offers:
id {
  value: "20141008-182754-3676708780-5050-27608-49558"
}
framework_id {
  value: "20141008-182754-3676708780-5050-27608-0002"
}
slave_id {
  value: "20141008-054505-3676708780-5050-21219-3"
}
hostname: "ip-172-31-45-108.eu-west-1.compute.internal"
resources {
  name: "cpus"
  type: SCALAR
  scalar {
    value: 2.0
  }
  role: "*"
}
resources {
  name: "mem"
  type: SCALAR
  scalar {
    value: 6489.0
  }
  role: "*"
}
resources {
  name: "disk"
  type: SCALAR
  scalar {
    value: 5023.0
  }
  role: "*"
}
resources {
  name: "ports"
  type: RANGES
  ranges {
    range {
      begin: 31000
      end: 32000
    }
  }
  role: "*"
}

Building task list...
Calling SchedulerDriver.launchTasks...
#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007fa34b14923d, pid=20973, tid=140339180459776
#
# JRE version: OpenJDK Runtime Environment (7.0_65-b32) (build 1.7.0_65-b32)
# Java VM: OpenJDK 64-Bit Server VM (24.65-b04 mixed mode linux-amd64 compressed oops)
# Derivative: IcedTea 2.5.2
# Distribution: Ubuntu 14.04 LTS, package 7u65-2.5.2-3~14.04
# Problematic frame:
# C  [libmesos-0.21.0.so+0xbac23d]  JNIEnv_::CallBooleanMethod(_jobject*, _jmethodID*, ...)+0x7d
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# /home/ubuntu/mesos-compat/hs_err_pid20973.log
#
# If you would like to submit a bug report, please include
# instructions on how to reproduce the bug and visit:
#   http://icedtea.classpath.org/bugzilla
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#
^CAborted (core dumped)