No rule to make target `/usr/include/leveldb/db.h'

[Rodrigo Naranjo]

Hello,

I'm new to caffe and I'm having some troubles installing caffe-segnet it on a remote machine. I ran make all, make test, make pycaffe and make distribute without any problems. The remote machine uses PBS (Portable Batch System) in order to schedule jobs and we are supposed to allocate resources through a bash script before sending them to a queue. When I run make runtest normally on the bash shell it throws the following error:

.build_release/tools/caffe
caffe: command line brew
usage: caffe <command> <args>


commands:
  train           train or finetune a model
  test            score a model
  device_query    show GPU diagnostic information
  time            benchmark model execution time


  Flags from tools/caffe.cpp:
    -gpu (Run in GPU mode on given device ID.) type: int32 default: -1
    -iterations (The number of iterations to run.) type: int32 default: 50
    -model (The model definition protocol buffer text file..) type: string
      default: ""
    -snapshot (Optional; the snapshot solver state to resume training.)
      type: string default: ""
    -solver (The solver definition protocol buffer text file.) type: string
      default: ""
    -weights (Optional; the pretrained weights to initialize finetuning. Cannot
      be set simultaneously with snapshot.) type: string default: ""
.build_release/test/test_all.testbin 0 --gtest_shuffle
Cuda number of devices: 0
Setting to use device 0
Current device id: 0
Note: Randomizing tests' orders with a seed of 16424 .
[==========] Running 1412 tests from 222 test cases.
[----------] Global test environment set-up.
[----------] 5 tests from MemoryDataLayerTest/3, where TypeParam = caffe::GPUDevice<double>
[ RUN      ] MemoryDataLayerTest/3.TestSetup
E0507 16:43:49.812096 21232 common.cpp:93] Cannot create Cublas handle. Cublas won't be available.
E0507 16:43:49.813638 21232 common.cpp:100] Cannot create Curand generator. Curand won't be available.
[       OK ] MemoryDataLayerTest/3.TestSetup (6 ms)
[ RUN      ] MemoryDataLayerTest/3.TestForward
[       OK ] MemoryDataLayerTest/3.TestForward (17 ms)
[ RUN      ] MemoryDataLayerTest/3.TestSetBatchSize
[       OK ] MemoryDataLayerTest/3.TestSetBatchSize (2 ms)
[ RUN      ] MemoryDataLayerTest/3.AddMatVectorDefaultTransform
[       OK ] MemoryDataLayerTest/3.AddMatVectorDefaultTransform (2 ms)
[ RUN      ] MemoryDataLayerTest/3.AddDatumVectorDefaultTransform
[       OK ] MemoryDataLayerTest/3.AddDatumVectorDefaultTransform (2 ms)
[----------] 5 tests from MemoryDataLayerTest/3 (30 ms total)


[----------] 5 tests from AdaGradSolverTest/2, where TypeParam = caffe::GPUDevice<float>
[ RUN      ] AdaGradSolverTest/2.TestAdaGradLeastSquaresUpdate
E0507 16:43:49.840687 21232 common.cpp:120] Curand not available. Skipping setting the curand seed.
F0507 16:43:49.844883 21232 syncedmem.cpp:57] Check failed: error == cudaSuccess (38 vs. 0)  no CUDA-capable device is detected
*** Check failure stack trace: ***
    @     0x7f3181773f8d  google::LogMessage::Fail()
    @     0x7f31817794c2  google::LogMessage::SendToLog()
    @     0x7f31817760e0  google::LogMessage::Flush()
    @     0x7f31817763cd  google::LogMessageFatal::~LogMessageFatal()
    @     0x7f317e05eba8  caffe::SyncedMemory::to_gpu()
    @     0x7f317e05e689  caffe::SyncedMemory::gpu_data()
    @     0x7f317df6811d  caffe::Blob<>::gpu_data()
    @     0x7f317e071514  caffe::InnerProductLayer<>::Forward_gpu()
    @     0x7f317e05dfba  caffe::Net<>::ForwardFromTo()
    @     0x7f317e05e306  caffe::Net<>::ForwardPrefilled()
    @           0x571745  caffe::GradientBasedSolverTest<>::ComputeLeastSquaresUpdate()
    @           0x572937  caffe::GradientBasedSolverTest<>::TestLeastSquaresUpdate()
    @           0x84718d  testing::internal::HandleExceptionsInMethodIfSupported<>()
    @           0x839081  testing::Test::Run()
    @           0x839167  testing::TestInfo::Run()
    @           0x8392a7  testing::TestCase::Run()
    @           0x83e1af  testing::internal::UnitTestImpl::RunAllTests()
    @           0x846d3d  testing::internal::HandleExceptionsInMethodIfSupported<>()
    @           0x8386aa  testing::UnitTest::Run()
    @           0x4d3bdf  main
    @     0x7f317d407d1d  __libc_start_main
    @           0x4d3939  (unknown)
make: *** [runtest] Aborted

I figured I'd have to send the "make runtest" through PBS by requesting usage of the GPU. However, when I send the job to the queue it fails with the following error: 

make: *** No rule to make target `/usr/include/leveldb/db.h', needed by `.build_release/cuda/src/caffe/test/test_im2col_kernel.o'.  Stop.

What could be causing this problem on my installation? Am I missing something on my Makefile? I have managed to install caffe-segnet using the CPU-only flag using the attached Makefile.config by uncommenting the "CPU_ONLY := 1" line, however training takes way too long without the support of a GPU.

Thanks in advance.

[Анна Куклева]

Hi! I think u should recompile caffe with USE_LEVELDB:=1, USE_LMDB:=1 flags. But I'm not sure because of Makefile.config.example C3D-v1.0 doesn't have these lines. Or u can try C3D-v1.1. For me this work

воскресенье, 7 мая 2017 г., 23:13:35 UTC+3 пользователь Rodrigo Naranjo написал: