Ruby on rails Rspec模型测试失败是由于Google Compute VM上的ENOMEM，而不是物理机器上的ENOMEM

在我的Google Compute服务器上用3.75 GB的RAM运行rspec模型测试时，我经常遇到内存分配错误，而在正常机器上用2 GB的RAM运行相同的测试时，我没有得到这些错误。我使用的命令是“rspec spec/models”，以便运行所有测试。当我运行任何其他类型的rspec测试时，都不会发生这种情况

在我的普通机器上，584个测试中只有2个失败（目前这是预期的行为）。但是，当我使用Google Compute server时，有148个故障发生，这是我从顶部获得的输出的一个示例：

Run options: include {:focus=>true}

All examples were filtered out; ignoring {:focus=>true}
....................F.FF.......................................................................................................................FFF.FF.F.............................................................................................................................F..F.F.F..FF.F...F.FF..F..F..F....F.FFF..FF.F.FF.FFFF.F.FFFFF.FF..F.FF.FFF...F.FFF...FFFF.FFFF.F.F.FF...............................................FF..F.F.FF.FF.FF.F...F.F.F..F.F.F.F......F.F...F.F.F.FF..FF.F.FFF..FF.FF.F..F..........FFFFFFFFFFFFFFFFFFFF
An error occurred in an after(:all) hook.
  Errno::ENOMEM: Cannot allocate memory - rm -rf /home/patrick/terrabase/imperial/files/test
  occurred at /home/patrick/.rvm/gems/ruby-1.9.3-p194@imperialhorizons/gems/activesupport-3.2.22/lib/active_support/core_ext/kernel/agnostics.rb:7:in ``'

.
An error occurred in an after(:all) hook.
  Errno::ENOMEM: Cannot allocate memory - rm -rf /home/patrick/terrabase/imperial/files/test
  occurred at /home/patrick/.rvm/gems/ruby-1.9.3-p194@imperialhorizons/gems/activesupport-3.2.22/lib/active_support/core_ext/kernel/agnostics.rb:7:in ``'

....
An error occurred in an after(:all) hook.
  Errno::ENOMEM: Cannot allocate memory - rm -rf /home/patrick/terrabase/imperial/files/test
  occurred at /home/patrick/.rvm/gems/ruby-1.9.3-p194@imperialhorizons/gems/activesupport-3.2.22/lib/active_support/core_ext/kernel/agnostics.rb:7:in ``'

.
An error occurred in an after(:all) hook.
  Errno::ENOMEM: Cannot allocate memory - rm -rf /home/patrick/terrabase/imperial/files/test
  occurred at /home/patrick/.rvm/gems/ruby-1.9.3-p194@imperialhorizons/gems/activesupport-3.2.22/lib/active_support/core_ext/kernel/agnostics.rb:7:in ``'

.

…这种情况一直持续到底部的“失败”部分：

142) LibraryBarcodeImport#process_row update existing student if the cid matches should set the imported count to 1
     Failure/Error: let(:library_barcode_import) { FactoryGirl.build(:library_barcode_import, log_text: "", imported_count: 0) }
     Errno::ENOMEM:
       Cannot allocate memory - file -b --mime '/home/patrick/terrabase/imperial/spec/testfiles/barcodes.xlsx'
     # ./spec/models/library_barcode_import_spec.rb:57:in `block (3 levels) in <top (required)>'
     # ./spec/models/library_barcode_import_spec.rb:103:in `block (4 levels) in <top (required)>'

…第103行在“before”语句中：

before do
    library_barcode_import.process_row(spreadsheet_row, 1, checking )
end

软件：

• Ruby 1.9.3-p194
• 轨道3.2.22
• Rspec 2.13.1

虚拟机规格：

• n1-standard-1（1个vCPU，3.75 GB内存）
• CPU平台：英特尔桑迪桥
• 操作系统：Ubuntu 14.04 64位，内核：3.19.0-43-generic
• 硬盘：10GB

物理机器规格：

• 东芝卫星专业A300
• CPU：英特尔奔腾双CPU T2390@1.86 GHz
• 操作系统：Linux Mint 17.3 64位

从“free-m”的输出可以看出，VM上不缺少可用的RAM：

             total       used       free     shared    buffers     cached
Mem:          3700        947       2753         14         80        531
-/+ buffers/cache:        336       3364
Swap:            0          0          0

“dmesg | grep-i virtual”的输出显示服务器依赖于KVM：

[    0.000000] Booting paravirtualized kernel on KVM

在谷歌上搜索时，我没有发现任何对处理这个问题有帮助的方法。有人知道这可能是什么原因吗

[    0.000000] Booting paravirtualized kernel on KVM