Update random slicing configurations (17fdd151) · Commits · David Auer / gekkofs

include/global/random_slicing/Disk.hpp

+7 −7

Original line number	Diff line number	Diff line
		@@ -36,7 +36,7 @@ public:
		* @param hostname TODO(dauer)
		* @param uri TODO(dauer)
		*/
		Disk(int64_t id, int64_t capacity, std::string hostname, std::string uri);
		Disk(uint64_t id, uint64_t capacity, std::string hostname, std::string uri);

		/**
		* copy constructor
		@@ -61,7 +61,7 @@ public:
		*
		* @return ID of the disk. Has to be unique all over this library.
		*/
		int64_t
		uint64_t
		getId() const;

		/**
		@@ -70,7 +70,7 @@ public:
		* @param id ID of the disk. Has to be unique all over this library.
		*/
		void
		setId(int64_t id);
		setId(uint64_t id);

		/**
		* Get the Capacity of the Disk in bytes. (This can also be in any
		@@ -80,7 +80,7 @@ public:
		* other scale, but has to be same for ExtentSize of
		* Distributor)
		*/
		int64_t
		uint64_t
		getCapacity() const;

		/**
		@@ -92,7 +92,7 @@ public:
		* Distributor)
		*/
		void
		setCapacity(int64_t capacity);
		setCapacity(uint64_t capacity);

		/**
		*/
		@@ -121,13 +121,13 @@ private:
		/**
		* ID of the disk. Has to be unique all over this library.
		*/
		int64_t id;
		uint64_t id;

		/**
		* Capacity of the Disk in bytes. (This can also be in any other
		* scale, but has to be same for ExtentSize of Distributor)
		*/
		int64_t capacity;
		uint64_t capacity;

		/**
		* String to be used by developer instantiating this disk. May not be

include/global/random_slicing/DistRandSlice.hpp

+1 −1

Original line number	Diff line number	Diff line
		@@ -112,7 +112,7 @@ public:
		* TODO(dauer): Decide & document freeing of memory
		*/
		void
		removeDisks(std::vector<int64_t> disk_ids);
		removeDisks(std::vector<uint64_t> disk_ids);

		/**
		* @see Distributor::getDisks

src/daemon/relocation/config_manager.cpp

+74 −34

Original line number	Diff line number	Diff line
		@@ -23,10 +23,12 @@
		#include <global/rpc/rpc_types.hpp>
		#include <global/rpc/distributor.hpp>

		#include <algorithm>
		#include <cassert>
		#include <filesystem>
		#include <iostream>
		#include <list>
		#include <set>

		#include <fmt/format.h>

		@@ -43,13 +45,9 @@ void
		calculate_random_slicing() {
		using gkfs::rpc::host_t;

		auto hosts = read_hosts_file();
		// use id only for new configurations!
		auto hostsfile_hosts = read_hosts_file();
		list<VDRIVE::Disk*> disks;
		for(const auto& [id, host] : hosts) {
		const auto& [hostname, uri] = host;
		disks.emplace_back(new VDRIVE::Disk(id, 1, hostname, uri));
		cout << uri << "\n";
		}
		auto filename = GKFS_DATA->hosts_file() + ".rsconfig";
		auto dist = std::make_shared<VDRIVE::DistRandSlice>();
		if(std::filesystem::exists(filename)) {
		@@ -65,40 +63,82 @@ calculate_random_slicing() {
		gkfs::rpc::RandomSlicingDistributor rsdist(dist);

		// get vector of all existing rs hosts
		auto all_rs_hosts = rsdist.locate_directory_metadata("");
		vector<host_t> all_hostsfile_hosts{};
		for(const auto& host : hosts) {
		all_hostsfile_hosts.push_back(host.first);
		auto rs_hosts = rsdist.get_hosts_map();
		uint64_t max_host_id = 0;
		std::set<std::string> rs_uris{};
		for(const auto& [id, host] : rs_hosts) {
		max_host_id = std::max(max_host_id, id);
		rs_uris.insert(host.second);
		}

		std::vector<std::string> gone_uris{};
		std::vector<uint64_t> gone_ids{};
		std::map<std::string, std::string> new_uris_hostname{};
		std::set<std::string> hostsfile_uris{};

		std::string new_uris_str{};
		std::string gone_uris_str{};

		for(const auto& [dont_care, host] : hostsfile_hosts) {
		const auto& [hostname, uri] = host;
		hostsfile_uris.insert(uri);
		if(rs_uris.count(uri) == 0) {
		new_uris_hostname[uri] = hostname;
		new_uris_str += "\n" + uri;
		}
		}

		// compare with read hosts
		vector<host_t> gone_hosts{};
		vector<host_t> new_hosts{};
		for(const auto& [id, host] : rs_hosts) {
		const auto& [hostname, uri] = host;
		if(hostsfile_uris.count(uri) == 0) {
		gone_uris.push_back(uri);
		gone_ids.push_back(id);
		gone_uris_str += "\n" + uri;
		}
		}

		// lists need to be sorted for set_difference
		std::sort(all_rs_hosts.begin(), all_rs_hosts.end());
		std::sort(all_hostsfile_hosts.begin(), all_hostsfile_hosts.end());

		GKFS_DATA->spdlogger()->info("Hosts to be added: {}", new_uris_str);
		GKFS_DATA->spdlogger()->info("Hosts to be removed: {}", gone_uris_str);

		/*
		// gone = rs_hosts \ hostsfile_hosts
		std::set_difference(all_rs_hosts.begin(), all_rs_hosts.end(),
		all_hostsfile_hosts.begin(),
		all_hostsfile_hosts.end(),
		std::inserter(gone_hosts, gone_hosts.begin()));
		std::inserter(gone_uris,
		gone_uris.begin()));

		// new = hostsfile_hosts \ rs_hosts
		std::set_difference(all_hostsfile_hosts.begin(),
		all_hostsfile_hosts.end(), all_rs_hosts.begin(),
		all_rs_hosts.end(),
		std::inserter(new_hosts, new_hosts.begin()));
		all_hostsfile_hosts.end(),
		all_rs_hosts.begin(), all_rs_hosts.end(), std::inserter(new_uris,
		new_uris.begin()));
		*/

		// remove gone hosts
		if(gone_ids.size() > 0) {
		dist->removeDisks(gone_ids);
		}
		if(new_uris_hostname.size() > 0) {
		// add new hosts
		// write to file
		// be done with it
		for(const auto& [uri, hostname] : new_uris_hostname) {
		disks.emplace_back(
		new VDRIVE::Disk(++max_host_id, 1, hostname, uri));
		}
		dist->addDisks(&disks);
		}
		} else {
		GKFS_DATA->spdlogger()->info(
		"No existing random slicing configuration found, "
		"creating from scratch with {} hosts.",
		hosts.size());
		hostsfile_hosts.size());

		for(const auto& [id, host] : hostsfile_hosts) {
		const auto& [hostname, uri] = host;
		disks.emplace_back(new VDRIVE::Disk(id, 1, hostname, uri));
		cout << uri << "\n";
		}
		dist->setConfiguration(&disks, 1, 1);
		}
		for(auto disk : disks) {
		@@ -110,7 +150,7 @@ calculate_random_slicing() {
		// Just test reading here
		VDRIVE::DistRandSlice dist_read{};
		dist_read.from_file(filename);
		dist_read.to_file(filename + ".reexport");
		dist_read.to_file(filename + ".debug-reexport");
		}

		unordered_map<uint64_t, pair<string, string>>

src/global/random_slicing/Disk.cpp

+5 −5

Original line number	Diff line number	Diff line
		@@ -16,7 +16,7 @@
		using namespace VDRIVE;
		using namespace std;

		Disk::Disk(int64_t id, int64_t capacity, std::string hostname,
		Disk::Disk(uint64_t id, uint64_t capacity, std::string hostname,
		std::string uri) {
		this->id = id;
		this->capacity = capacity;
		@@ -81,22 +81,22 @@ Disk::setUri(std::string uri) {
		this->uri = uri;
		}

		int64_t
		uint64_t
		Disk::getId() const {
		return this->id;
		}

		void
		Disk::setId(int64_t id) {
		Disk::setId(uint64_t id) {
		this->id = id;
		}

		int64_t
		uint64_t
		Disk::getCapacity() const {
		return this->capacity;
		}

		void
		Disk::setCapacity(int64_t capacity) {
		Disk::setCapacity(uint64_t capacity) {
		this->capacity = capacity;
		}

src/global/random_slicing/DistRandSlice.cpp

+1 −1

Original line number	Diff line number	Diff line
		@@ -126,7 +126,7 @@ DistRandSlice::addDisks(std::list<Disk> disks) {

		// Added for GekkoFS
		void
		DistRandSlice::removeDisks(std::vector<int64_t> disk_ids) {
		DistRandSlice::removeDisks(std::vector<uint64_t> disk_ids) {
		// part id -> num blocks to remove
		std::unordered_map<uint64_t, uint64_t> old_partitions;
		for(const auto partition : *m_partitions) {