Skip to content

Commit

Permalink
Scraper greylisting machinery (Part 1)
Browse files Browse the repository at this point in the history
  • Loading branch information
@jamescowens
jamescowens committed Oct 3, 2024
1 parent 867ec98 commit 715dba6
Show file tree
Hide file tree
Showing 2 changed files with 95 additions and 30 deletions.
5 changes: 5 additions & 0 deletions src/gridcoin/scraper/fwd.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -159,6 +159,11 @@ struct ConvergedManifest
*/
std::vector<std::string> vExcludedProjects;

//!
//! \brief The list of projects that have been greylisted.
//!
std::vector<std::string> vGreylistedProjects;

/** Populates the part pointers map in the convergence */
bool PopulateConvergedManifestPartPtrsMap();

Expand Down
120 changes: 90 additions & 30 deletions src/gridcoin/scraper/scraper.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1567,7 +1567,7 @@ void Scraper(bool bSingleShot)
uiInterface.NotifyScraperEvent(scrapereventtypes::Stats, CT_UPDATING, {});

// Get a read-only view of the current project whitelist:
const WhitelistSnapshot projectWhitelist = GetWhitelist().Snapshot();
const WhitelistSnapshot projectWhitelist = GetWhitelist().Snapshot(GRC::ProjectEntry::ProjectFilterFlag::ALL_BUT_DELETED);

// Delete manifest entries not on whitelist. Take a lock on cs_StructScraperFileManifest for this.
{
Expand Down Expand Up @@ -3630,7 +3630,8 @@ bool ProcessProjectStatsFromStreamByCPID(const std::string& project, boostio::fi

// At the individual (byCPIDbyProject) level the AvgRAC is the same as the RAC.
statsentry.statsvalue.dAvgRAC = statsentry.statsvalue.dRAC;
// Mag is dealt with on the second pass... so is left at 0.0 on the first pass.
// Mag is dealt with on the second pass, so is set to 0.0 on the first pass.
statsentry.statsvalue.dMag = 0.0;

statsentry.statskey.objecttype = statsobjecttype::byCPIDbyProject;
statsentry.statskey.objectID = project + "," + cpid;
Expand All @@ -3648,20 +3649,24 @@ bool ProcessProjectStatsFromStreamByCPID(const std::string& project, boostio::fi
// The mScraperStats here is scoped to only this project so we do not need project filtering here.
ScraperStats::iterator entry;

for (auto const& entry : mScraperStats)
{
ScraperObjectStats statsentry;
// Statistics tracked for greylisted projects have zero project magnitude, so no need to go through
// and update the CPID level mags. They are all zero if project magnitude is zero.
if (projectmag > 0) {
for (auto const& entry : mScraperStats)
{
ScraperObjectStats statsentry;

statsentry.statskey = entry.first;
statsentry.statsvalue.dTC = entry.second.statsvalue.dTC;
statsentry.statsvalue.dRAT = entry.second.statsvalue.dRAT;
statsentry.statsvalue.dRAC = entry.second.statsvalue.dRAC;
// As per the above the individual (byCPIDbyProject) level the AvgRAC is the same as the RAC.
statsentry.statsvalue.dAvgRAC = entry.second.statsvalue.dAvgRAC;
statsentry.statsvalue.dMag = MagRound(entry.second.statsvalue.dRAC / dProjectRAC * projectmag);
statsentry.statskey = entry.first;
statsentry.statsvalue.dTC = entry.second.statsvalue.dTC;
statsentry.statsvalue.dRAT = entry.second.statsvalue.dRAT;
statsentry.statsvalue.dRAC = entry.second.statsvalue.dRAC;
// As per the above the individual (byCPIDbyProject) level the AvgRAC is the same as the RAC.
statsentry.statsvalue.dAvgRAC = entry.second.statsvalue.dAvgRAC;
statsentry.statsvalue.dMag = MagRound(entry.second.statsvalue.dRAC / dProjectRAC * projectmag);

// Update map entry with the magnitude.
mScraperStats[statsentry.statskey] = statsentry;
// Update map entry with the magnitude.
mScraperStats[statsentry.statskey] = statsentry;
}
}

// Due to rounding to MAG_ROUND, the actual total project magnitude will not be exactly projectmag,
Expand Down Expand Up @@ -3796,17 +3801,21 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsByCurrentFileManifestState()
{
_log(logattribute::INFO, "GetScraperStatsByCurrentFileManifestState", "Beginning stats processing.");

// Get a read-only view of the current project greylist
const WhitelistSnapshot greylist = GetWhitelist().Snapshot(GRC::ProjectEntry::ProjectFilterFlag::GREYLISTED);

// Enumerate the count of active projects from the file manifest. Since the manifest is
// constructed starting with the whitelist, and then using only the current files, this
// will always be less than or equal to the whitelist count from whitelist.
unsigned int nActiveProjects = 0;
{
LOCK(cs_StructScraperFileManifest);

for (auto const& entry : StructScraperFileManifest.mScraperFileManifest)
{
//
if (entry.second.current && !entry.second.excludefromcsmanifest) nActiveProjects++;
for (const auto& entry : StructScraperFileManifest.mScraperFileManifest) {
// Count as active if current, not marked as to be excluded, and also not greylisted.
if (entry.second.current
&& !entry.second.excludefromcsmanifest
&& !greylist.Contains(entry.second.project)) nActiveProjects++;
}
}
double dMagnitudePerProject = NETWORK_MAGNITUDE / nActiveProjects;
Expand All @@ -3831,7 +3840,12 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsByCurrentFileManifestState()
_log(logattribute::INFO, "GetScraperStatsByCurrentFileManifestState",
"Processing stats for project: " + project);

LoadProjectFileToStatsByCPID(project, file, dMagnitudePerProject, mProjectScraperStats);
if (!greylist.Contains(entry.second.project)) {
LoadProjectFileToStatsByCPID(project, file, dMagnitudePerProject, mProjectScraperStats);
} else {
// Project magnitude for a greylisted project is zero.
LoadProjectFileToStatsByCPID(project, file, 0.0, mProjectScraperStats);
}

// Insert into overall map.
for (auto const& entry2 : mProjectScraperStats)
Expand Down Expand Up @@ -3868,6 +3882,9 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsByConvergedManifest(const Converge
{
_log(logattribute::INFO, "GetScraperStatsByConvergedManifest", "Beginning stats processing.");

// Get a read-only view of the current project greylist
const WhitelistSnapshot greylist = GetWhitelist().Snapshot(GRC::ProjectEntry::ProjectFilterFlag::GREYLISTED);

ScraperStatsAndVerifiedBeacons stats_and_verified_beacons;

// Enumerate the count of active projects from the dummy converged manifest. One of the parts
Expand Down Expand Up @@ -3897,6 +3914,14 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsByConvergedManifest(const Converge
stats_and_verified_beacons.mVerifiedMap = VerifiedBeaconMap;

unsigned int nActiveProjects = StructConvergedManifest.ConvergedManifestPartPtrsMap.size() - exclude_parts_from_count;

// If a project part is greylisted, do not count it as an active project, even though stats have been collected.
for (const auto& project : StructConvergedManifest.ConvergedManifestPartPtrsMap) {
if (greylist.Contains(project.first)) {
--nActiveProjects;
}
}

_log(logattribute::INFO, "GetScraperStatsByConvergedManifest",
"Number of active projects in converged manifest = " + ToString(nActiveProjects));

Expand All @@ -3916,7 +3941,12 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsByConvergedManifest(const Converge
{
_log(logattribute::INFO, "GetScraperStatsByConvergedManifest", "Processing stats for project: " + project);

LoadProjectObjectToStatsByCPID(project, entry->second->data, dMagnitudePerProject, mProjectScraperStats);
if (!greylist.Contains(project)) {
LoadProjectObjectToStatsByCPID(project, entry->second->data, dMagnitudePerProject, mProjectScraperStats);
} else {
// Project magnitude for a greylisted project is zero.
LoadProjectObjectToStatsByCPID(project, entry->second->data, 0.0, mProjectScraperStats);
}

// Insert into overall map.
for (auto const& entry2 : mProjectScraperStats)
Expand All @@ -3939,6 +3969,9 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsFromSingleManifest(CScraperManifes
{
_log(logattribute::INFO, "GetScraperStatsFromSingleManifest", "Beginning stats processing.");

// Get a read-only view of the current project greylist
const WhitelistSnapshot greylist = GetWhitelist().Snapshot(GRC::ProjectEntry::ProjectFilterFlag::GREYLISTED);

// Create a dummy converged manifest and fill out the dummy ConvergedManifest structure from the provided
// manifest.
ConvergedManifest StructDummyConvergedManifest(manifest);
Expand Down Expand Up @@ -3973,6 +4006,14 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsFromSingleManifest(CScraperManifes

unsigned int nActiveProjects = StructDummyConvergedManifest.ConvergedManifestPartPtrsMap.size()
- exclude_parts_from_count;

// If a project part is greylisted, do not count it as an active project, even though stats have been collected.
for (const auto& project : StructDummyConvergedManifest.ConvergedManifestPartPtrsMap) {
if (greylist.Contains(project.first)) {
--nActiveProjects;
}
}

_log(logattribute::INFO, "GetScraperStatsFromSingleManifest",
"Number of active projects in converged manifest = " + ToString(nActiveProjects));

Expand All @@ -3989,7 +4030,12 @@ ScraperStatsAndVerifiedBeacons GetScraperStatsFromSingleManifest(CScraperManifes
{
_log(logattribute::INFO, "GetScraperStatsFromSingleManifest", "Processing stats for project: " + project);

LoadProjectObjectToStatsByCPID(project, entry->second->data, dMagnitudePerProject, mProjectScraperStats);
if (!greylist.Contains(project)) {
LoadProjectObjectToStatsByCPID(project, entry->second->data, dMagnitudePerProject, mProjectScraperStats);
} else {
// Project magnitude for a greylisted project is zero.
LoadProjectObjectToStatsByCPID(project, entry->second->data, 0.0, mProjectScraperStats);
}

// Insert into overall map.
stats_and_verified_beacons.mScraperStats.insert(mProjectScraperStats.begin(), mProjectScraperStats.end());
Expand Down Expand Up @@ -4852,7 +4898,7 @@ bool ScraperConstructConvergedManifest(ConvergedManifest& StructConvergedManifes

// Get a read-only view of the current project whitelist to fill out the
// excluded projects vector later on:
const WhitelistSnapshot projectWhitelist = GetWhitelist().Snapshot();
const WhitelistSnapshot projectWhitelist = GetWhitelist().Snapshot(GRC::ProjectEntry::ProjectFilterFlag::ALL_BUT_DELETED);

if (bConvergenceSuccessful)
{
Expand Down Expand Up @@ -4881,6 +4927,12 @@ bool ScraperConstructConvergedManifest(ConvergedManifest& StructConvergedManifes
// to try and recover project by project.
for (const auto& iProjects : projectWhitelist)
{
// If project is greylisted, push project name to greylist vector.
if (iProjects.m_status == GRC::ProjectEntryStatus::MAN_GREYLISTED
|| iProjects.m_status == GRC::ProjectEntryStatus::AUTO_GREYLISTED) {
StructConvergedManifest.vGreylistedProjects.push_back(iProjects.m_name);
}

if (StructConvergedManifest.ConvergedManifestPartPtrsMap.find(iProjects.m_name)
== StructConvergedManifest.ConvergedManifestPartPtrsMap.end())
{
Expand Down Expand Up @@ -4964,6 +5016,12 @@ bool ScraperConstructConvergedManifestByProject(const WhitelistSnapshot& project

for (const auto& iWhitelistProject : projectWhitelist)
{
// If project is greylisted, push project name to greylist vector for later use.
if (iWhitelistProject.m_status == GRC::ProjectEntryStatus::MAN_GREYLISTED
||iWhitelistProject.m_status == GRC::ProjectEntryStatus::AUTO_GREYLISTED) {
StructConvergedManifest.vGreylistedProjects.push_back(iWhitelistProject.m_name);
}

// Do a map for unique ProjectObject times ordered by descending time then content hash. Note that for Project
// Objects (Parts), the content hash is the object hash. We also need the consensus block here, because we are
// "composing" the manifest by parts, so we will need to choose the latest consensus block by manifest time. This
Expand Down Expand Up @@ -5135,8 +5193,10 @@ bool ScraperConstructConvergedManifestByProject(const WhitelistSnapshot& project

auto convergence_by_project_ratio = [](){ LOCK(cs_ScraperGlobals); return CONVERGENCE_BY_PROJECT_RATIO; };

// If we meet the rule of CONVERGENCE_BY_PROJECT_RATIO, then proceed to fill out the rest of the map.
if ((double)iCountSuccessfulConvergedProjects / (double)projectWhitelist.size() >= convergence_by_project_ratio())
// If we meet the rule of CONVERGENCE_BY_PROJECT_RATIO, then proceed to fill out the rest of the map. Note that the greylisted
// projects are excluded from the count in the denominator as it is not expected to necessarily achieve convergence on those.
if ((double)iCountSuccessfulConvergedProjects /
(double)(projectWhitelist.size() - StructConvergedManifest.vGreylistedProjects.size()) >= convergence_by_project_ratio())
{
AppCacheSection mScrapers = GetScrapersCache();

Expand Down Expand Up @@ -6072,13 +6132,13 @@ UniValue testnewsb(const UniValue& params, bool fHelp)
{
LOCK(cs_ConvergedScraperStatsCache);

if (!ConvergedScraperStatsCache.NewFormatSuperblock.WellFormed())
{
UniValue error(UniValue::VOBJ);
error.pushKV("Error:", "Wait until a convergence is formed.");
if (!ConvergedScraperStatsCache.NewFormatSuperblock.WellFormed())
{
UniValue error(UniValue::VOBJ);
error.pushKV("Error:", "Wait until a convergence is formed.");

return error;
}
return error;
}

_log(logattribute::INFO, "testnewsb",
"Size of the PastConvergences map = " + ToString(ConvergedScraperStatsCache.PastConvergences.size()));
Expand Down

0 comments on commit 715dba6

Please sign in to comment.