zfsbackupcleaner/main.cpp

#include <boost/date_time/posix_time/posix_time_config.hpp>
#include <boost/date_time/posix_time/posix_time_types.hpp>
#include <boost/date_time/posix_time/time_formatters.hpp>
#include <boost/date_time/posix_time/time_parsers.hpp>
#include <iostream>

#include <boost/process.hpp>
#include <boost/process/v2/stdio.hpp>
#include <boost/process/v2/environment.hpp>
#include <boost/asio.hpp>

#include <boost/date_time/posix_time/posix_time.hpp>
#include <set>

namespace bp = boost::process;
namespace asio = boost::asio;

bool dry_run = true;

template<int keep_every_x_days,int till_y_day>std::set<boost::posix_time::ptime>::reverse_iterator handle_backups_after(std::string & snapshotbase, std::set<boost::posix_time::ptime> & times, std::set<boost::posix_time::ptime>::reverse_iterator ct, boost::posix_time::ptime now)
{
    if(ct == times.rend())
        return ct;

    boost::posix_time::time_duration dur;
    boost::posix_time::ptime last_kept_time;
    
    while( ct != times.rend() && (dur = now - *ct).hours() <= till_y_day * 24)
    {
        last_kept_time = *ct;
        std::cout << "Keeping Backup " << last_kept_time << " of this day." << std::endl << std::endl;
        ct++;

        while( ct != times.rend() && (dur = last_kept_time - *ct).hours() < keep_every_x_days * 24)
        {
            if(ct == --times.rend())
            {
                std::cout << "Keeping last Backup " << *ct << " in this backup-set."  << std::endl << std::endl;
                return ct++;
            }

            std::cout << "Removing Backup " << *ct << " because we have already one in the relevant set of keeping every " << keep_every_x_days << " till " << till_y_day << " days."  << std::endl;

            std::string timestring = boost::posix_time::to_iso_extended_string(*ct);
            timestring[10]='_';
            timestring[13]='-';
            timestring[16]='-';
            timestring=timestring.substr(0, 19);

            asio::io_context ctx;
            asio::readable_pipe out(ctx);

            std::vector<std::string> args;
            args.push_back("destroy");
            args.push_back("-v");
            if(dry_run)
                args.push_back("-n");
            args.push_back(snapshotbase + "@" + timestring);

            bp::process proc(ctx, bp::environment::find_executable("zfs"), args, bp::process_stdio{{/* in to default*/}, out, {/* err to default */}});

            std::string output;

            boost::system::error_code ec;
            asio::read(out, asio::dynamic_buffer(output), ec);
            assert(!ec || (ec==asio::error::eof));
            proc.wait();

            std::cout << output << std::endl;

            ct++;
        }
    }

    return ct;
}

std::set<boost::posix_time::ptime>::reverse_iterator handle_backups_after_one_day(std::string & snapshotbase, std::set<boost::posix_time::ptime> & times, std::set<boost::posix_time::ptime>::reverse_iterator ct, boost::posix_time::ptime now)
{
    // after one day, we keep one backup per day

    return handle_backups_after<1,30>(snapshotbase, times, ct, now);

    /*if(ct == times.rend())
        return ct;

    boost::posix_time::time_duration dur;
    boost::posix_time::ptime last_kept_time;
    
    while( ct != times.rend() && (dur = now - *ct).hours() <= 30 * 24)
    {
        last_kept_time = *ct;
        std::cout << "Keeping Backup " << last_kept_time << " of this day." << std::endl;
        ct++;

        while( ct != times.rend() && (dur = last_kept_time - *ct).hours() < 24)
        {
            std::cout << "Removing Backup " << *ct << " because we have already one in 24 hours." << std::endl;
             ct++;
        }
    }

    return ct;*/
}

std::set<boost::posix_time::ptime>::reverse_iterator handle_backups_after_30_days(std::string & snapshotbase, std::set<boost::posix_time::ptime> & times, std::set<boost::posix_time::ptime>::reverse_iterator ct, boost::posix_time::ptime now)
{
    // after 30 days, we keep one backup every 7 days

    return handle_backups_after<7,365>(snapshotbase, times, ct, now);
}

std::set<boost::posix_time::ptime>::reverse_iterator handle_backups_after_365_days(std::string & snapshotbase, std::set<boost::posix_time::ptime> & times, std::set<boost::posix_time::ptime>::reverse_iterator ct, boost::posix_time::ptime now)
{
    // after 365 days, we keep one backup every 30 days

    return handle_backups_after<30, 365*100>(snapshotbase, times, ct, now);
}


int main(int, char**){
    std::cout << "Hello, from zfsbackupcleaner!\n";

    asio::io_context ctx;
    asio::readable_pipe out(ctx);

    bp::process proc(ctx, bp::environment::find_executable("zfs"), {"list", "-t", "snapshot", "/backup/workstation"}, bp::process_stdio{{/* in to default*/}, out, {/* err to default */}});

    std::string output;

    boost::system::error_code ec;
    asio::read(out, asio::dynamic_buffer(output), ec);
    assert(!ec || (ec==asio::error::eof));
    proc.wait();

    //std::cout << "OUTPUT" << output << "END" << std::endl;

    std::stringstream lines(output);

    std::set<boost::posix_time::ptime> times;

    
    std::string snapshotbase;

    for(std::string line; std::getline(lines, line);)
    {
        std::string::size_type pos = line.find(' ');
        line = line.substr(0,pos);
        //std::cout << "X: " << line << std::endl;

        if(line == "NAME")
            continue;

        pos = line.find('@');
        snapshotbase = line.substr(0,pos);
        std::string timestring = line.substr(pos+1);
        timestring[10]=' ';
        timestring[13]=':';
        timestring[16]=':';
        boost::posix_time::ptime t(boost::posix_time::time_from_string(timestring));

        times.insert(t);
    }

    if(times.empty())
    {
        std::cout << "NO SNAPSHOTS FOUND — EXIT";
        return 0;
    }

    std::set<boost::posix_time::ptime>::reverse_iterator current_time = times.rbegin();

    // handle fist 24 hours!!!

    boost::posix_time::ptime now = boost::posix_time::second_clock::local_time();

    std::cout << "NOW " << now << std::endl;

    boost::posix_time::time_duration dur;
    
    while( (dur = now - *current_time).hours() <= 72)
    {
        std::cout << "Omitting Backup " << *current_time << " because it is too young to handle (" << dur.hours() << " hours old)." << std::endl;
        current_time++;
    }
    std::cout << std::endl;
    
    current_time = handle_backups_after_one_day(snapshotbase, times, current_time, now);
    current_time = handle_backups_after_30_days(snapshotbase, times, current_time, now);
    current_time = handle_backups_after_365_days(snapshotbase, times, current_time, now);

    return 0;
}