Commit 9479232e authored by Ramon Nou's avatar Ramon Nou Committed by Ramon Nou
Browse files

Implemented Guided Distributor (rebase master from GekkoFWD), Added Hostname in client logs

parent 4e873c61

CMakeLists.txt

+10 −0
Original line number Diff line number Diff line
@@ -141,6 +141,16 @@ add_definitions(-DLIBGKFS_LOG_MESSAGE_SIZE=${CLIENT_LOG_MESSAGE_SIZE}) 
message(STATUS "[gekkofs] Maximum log message size in the client library: ${CLIENT_LOG_MESSAGE_SIZE}")

mark_as_advanced(CLIENT_LOG_MESSAGE_SIZE)



option(USE_GUIDED "Use guided data distributor " OFF)

message(STATUS "[gekkofs] Guided data distributor: ${USE_GUIDED}")



set(USE_GUIDED_PATH "~/guided.txt" CACHE STRING "File Path for guided distributor")

set_property(CACHE USE_GUIDED_PATH PROPERTY STRINGS)

message(STATUS "[gekkofs] Guided data distributor input file path: ${USE_GUIDED_PATH}")



option(TRACE_GUIDED "Output at INFO level information for guided distributor generation: " OFF)

message(STATUS "[gekkofs] Generate log line at INFO level for guided distributor: ${TRACE_GUIDED}")



configure_file(include/global/cmake_configure.hpp.in include/global/cmake_configure.hpp)



# Imported target

README.md

+34 −0
Original line number Diff line number Diff line
@@ -239,6 +239,40 @@ can be provided to set the path to the log file, and the log module can be 
selected with the `GKFS_LOG_LEVEL={off,critical,err,warn,info,debug,trace}`

environment variable.



### Data distributors

The data distribution can be selected at compilation time, we have 2 distributors available:



## Simple Hash (Default)

Chunks are distributed randomly to the different GekkoFS servers.



## Guided Distributor

Guided distributor distributes chunks using a shared file with the next format:

`<path> <chunk_number> <host>`



Chunks not specified, are distributed using the Simple Hash distributor.



To generate such file we need to follow a first execution, using the next compilation options:

* `TRACE_GUIDED` ON

* `USE_GUIDED` OFF



This will enable a `INFO` level log at the clients offering several lines that can be used to generate the input file. 

In this stage, each node should generate a separated file this can be done in SLURM using the next line :

`srun -N 10 -n 320 --export="ALL" /bin/bash -c "export LIBGKFS_LOG_OUTPUT=${HOME}/test/GLOBAL.txt;LD_PRELOAD=${GKFS_PRLD} <app>"`



Then, use the `utils/generate.py` to create the output file. 

* `python utils/generate.py ~/test/GLOBAL.txt >> guided.txt`



This should work if the nodes are sorted in alphabetical order, which is the usual scenario.



```

Finally, enable the distributor using the next compilation flags:

* `TRACE_GUIDED` OFF

* `USE_GUIDED` ON

* `USE_GUIDED_PATH` `<path to guided.txt>`











### Acknowledgment

include/client/preload_context.hpp

+7 −1
Original line number Diff line number Diff line
@@ -20,6 +20,7 @@ 
#include <memory>

#include <vector>

#include <string>

#include <config.hpp>



#include <bitset>
@@ -92,7 +93,9 @@ private: 
    mutable std::mutex internal_fds_mutex_;

    bool internal_fds_must_relocate_;

    std::bitset<MAX_USER_FDS> protected_fds_;



#ifdef TRACE_GUIDED

    std::string hostname = "";

#endif

public:

    static PreloadContext* getInstance() {

        static PreloadContext instance;
@@ -167,6 +170,9 @@ public: 
    void protect_user_fds();



    void unprotect_user_fds();

#ifdef TRACE_GUIDED

    std::string get_hostname();

#endif

};



} // namespace preload

include/global/cmake_configure.hpp.in

+3 −0
Original line number Diff line number Diff line
@@ -16,5 +16,8 @@ 


#cmakedefine01 CREATE_CHECK_PARENTS

#cmakedefine01 LOG_SYSCALLS

#cmakedefine USE_GUIDED

#cmakedefine TRACE_GUIDED

#define GUIDED_PATH "@USE_GUIDED_PATH@"



#endif //FS_CMAKE_CONFIGURE_H

include/global/rpc/distributor.hpp

+23 −0
Original line number Diff line number Diff line
@@ -14,9 +14,12 @@ 
#ifndef GEKKOFS_RPC_DISTRIBUTOR_HPP

#define GEKKOFS_RPC_DISTRIBUTOR_HPP



#include "../include/config.hpp"

#include <vector>

#include <string>

#include <numeric>

#include <unordered_map>

#include <fstream>



namespace gkfs {

namespace rpc {
@@ -87,6 +90,26 @@ public: 
    std::vector<host_t> locate_directory_metadata(const std::string& path) const override;

};



class GuidedDistributor : public Distributor {

private:

    host_t localhost_;

    unsigned int hosts_size_;

    std::vector<host_t> all_hosts_;

    std::hash<std::string> str_hash;

    std::unordered_map< std::string, host_t > mapping;

public:

    GuidedDistributor(host_t localhost, unsigned int hosts_size);



    host_t localhost() const override;



    host_t locate_data(const std::string& path, const chunkid_t& chnk_id) const override;



    host_t locate_file_metadata(const std::string& path) const override;



    std::vector<host_t> locate_directory_metadata(const std::string& path) const override;

};





} // namespace rpc

} // namespace gkfs