[jira] [Created] (MESOS-6986) abort in DRFSorter::add

["Yvan Royon (JIRA)" <ji...@apache.org>]

Yvan Royon created MESOS-6986:
---------------------------------

             Summary: abort in DRFSorter::add
                 Key: MESOS-6986
                 URL: https://issues.apache.org/jira/browse/MESOS-6986
             Project: Mesos
          Issue Type: Bug
          Components: allocation
    Affects Versions: 1.0.1
         Environment: Mesosphere Enterprise DC/OS, CoreOS
            Reporter: Yvan Royon


My mesos-master process terminated on SIGABRT.
The CHECK failed in function {{DRFSorter::add}}:
https://github.com/apache/mesos/blob/master/src/master/allocator/sorter/drf/sorter.cpp#L74

It seems there is a condition during framework registration where names are lost?

We are using the mesos-go library ({{next}} branch), which uses the new HTTP API. The framework is custom Go code. The crash is hard to reliably reproduce.

{code}
mesos-master[90061]: F0119 01:07:57.426159 90086 sorter.cpp:73] Check failed: !contains(name)
mesos-master[90061]: *** Check failure stack trace: ***
mesos-master[90061]: @     0x7f960d9299fd  google::LogMessage::Fail()
mesos-master[90061]: @     0x7f960d92b82d  google::LogMessage::SendToLog()
mesos-master[90061]: @     0x7f960d9295ec  google::LogMessage::Flush()
mesos-master[90061]: @     0x7f960d92c129  google::LogMessageFatal::~LogMessageFatal()
mesos-master[90061]: @     0x7f960d03460d  mesos::internal::master::allocator::DRFSorter::add()
mesos-master[90061]: @     0x7f960d021177  mesos::internal::master::allocator::internal::HierarchicalAllocatorProcess::addFramework()
mesos-master[90061]: @     0x7f960d8b9381  process::ProcessManager::resume()
mesos-master[90061]: @     0x7f960d8b9687  _ZNSt6thread5_ImplISt12_Bind_simpleIFZN7process14ProcessManager12init_threadsEvEUt_vEEE6_M_runEv
mesos-master[90061]: @     0x7f960bf52d73  (unknown)
mesos-master[90061]: @     0x7f960b74f52c  (unknown)
mesos-master[90061]: @     0x7f960b49180d  (unknown)
systemd[1]: dcos-mesos-master.service: Main process exited, code=killed, status=6/ABRT
{code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)