[tor-commits] [metrics-tasks/master] Investigate lifetime of introduction points (#15513).
karsten at torproject.org
karsten at torproject.org
Tue Apr 14 15:43:21 UTC 2015
commit e0862169e12b9a687d5d05b6dd32688cfbfe05f7
Author: Karsten Loesing <karsten.loesing at gmx.net>
Date: Tue Apr 14 17:36:03 2015 +0200
Investigate lifetime of introduction points (#15513).
---
task-15513/.gitignore | 8 +
task-15513/README.md | 24 ++
task-15513/plot.R | 81 +++++++
task-15513/src/ParseDescriptors.java | 420 ++++++++++++++++++++++++++++++++++
4 files changed, 533 insertions(+)
diff --git a/task-15513/.gitignore b/task-15513/.gitignore
new file mode 100644
index 0000000..3f98e38
--- /dev/null
+++ b/task-15513/.gitignore
@@ -0,0 +1,8 @@
+.classpath
+.project
+descriptors/
+/bin
+*.csv
+lifetimes-verbose.txt
+Rplots.pdf
+
diff --git a/task-15513/README.md b/task-15513/README.md
new file mode 100644
index 0000000..f1bdd55
--- /dev/null
+++ b/task-15513/README.md
@@ -0,0 +1,24 @@
+Investigate lifetime of IPs on Hidden Services
+==============================================
+
+Extract hidden-service descriptors to descriptors/.
+
+Obtain commons-codec-1.6.jar and place it in /usr/share/java/.
+
+Compile and run the Java class:
+
+ $ javac -d bin/ -cp /usr/share/java/commons-codec-1.6.jar \
+ src/ParseDescriptors.java
+
+ $ java -cp bin/:/usr/share/java/commons-codec-1.6.jar ParseDescriptors
+
+Find verbose logs in lifetimes-verbose.txt.
+
+Make sure you have R 3.0.2 with ggplot2 0.9.3.1 installed.
+
+Plot graphs:
+
+ $ R --slave -f plot.R
+
+View output in Rplots.pdf
+
diff --git a/task-15513/plot.R b/task-15513/plot.R
new file mode 100644
index 0000000..f1ce9d3
--- /dev/null
+++ b/task-15513/plot.R
@@ -0,0 +1,81 @@
+require(ggplot2)
+require(reshape)
+require(scales)
+
+l <- read.csv("lifetimes.csv")
+ggplot(l, aes(x = lifetime_millis / 3600000, colour = service)) +
+ stat_ecdf() +
+ scale_x_continuous("\nLifetime in hours",
+ limits = c(0, max(l$lifetime_millis) / 3600000)) +
+ scale_y_continuous("Cumulative fraction\n", labels = percent) +
+ scale_colour_hue("Service") +
+ ggtitle("Lifetime of introduction points\n")
+
+l <- read.csv("published-descs.csv")
+ggplot(l, aes(x = descs, colour = service)) +
+ stat_ecdf() +
+ scale_x_continuous("\nNumber of descriptors",
+ limits = c(0, max(l$descs))) +
+ scale_y_continuous("Cumulative fraction\n", labels = percent) +
+ scale_colour_hue("Service") +
+ ggtitle(paste("Number of descriptors published per hour,\n",
+ "including descriptor replicas\n", sep = ""))
+
+l <- read.csv("established-intros.csv")
+ggplot(l, aes(x = intros, colour = service)) +
+ stat_ecdf() +
+ scale_x_continuous("\nNumber of introduction points",
+ limits = c(0, max(l$intros))) +
+ scale_y_continuous("Cumulative fraction\n", labels = percent) +
+ scale_colour_hue("Service") +
+ ggtitle("Number of distinct introduction points used per hour\n")
+
+l <- read.csv("intros-per-desc.csv")
+ggplot(l, aes(x = introsperdesc, colour = service)) +
+ stat_ecdf() +
+ scale_x_continuous("\nNumber of introduction points per descriptor",
+ limits = c(0, max(l$introsperdesc)),
+ breaks = seq(0, max(l$introsperdesc), 2)) +
+ scale_y_continuous("Cumulative fraction\n", labels = percent) +
+ scale_colour_hue("Service") +
+ ggtitle("Number of introduction points per descriptor\n")
+
+l <- read.csv("intros-per-relay.csv")
+data <- rbind(
+ data.frame(x = sort(l[l$service == "4cjw6cwpeaeppfqz" , 2]),
+ y = (1:length(l[l$service == "4cjw6cwpeaeppfqz" , 2])),
+ colour = "4cjw6cwpeaeppfqz"),
+ data.frame(x = sort(l[l$service == "agorahooawayyfoe" , 2]),
+ y = (1:length(l[l$service == "agorahooawayyfoe" , 2])),
+ colour = "agorahooawayyfoe"),
+ data.frame(x = sort(l[l$service == "facebookcorewwwi" , 2]),
+ y = (1:length(l[l$service == "facebookcorewwwi" , 2])),
+ colour = "facebookcorewwwi"),
+ data.frame(x = sort(l[l$service == "kpvz7kpmcmne52qf" , 2]),
+ y = (1:length(l[l$service == "kpvz7kpmcmne52qf" , 2])),
+ colour = "kpvz7kpmcmne52qf"),
+ data.frame(x = sort(l[l$service == "xdndo2okt43cjx44" , 2]),
+ y = (1:length(l[l$service == "xdndo2okt43cjx44" , 2])),
+ colour = "xdndo2okt43cjx44"))
+
+ggplot(data) +
+ geom_step(aes(x = x, y = y, colour = as.factor(colour))) +
+ scale_x_continuous("\nIntroduction points established on same relay",
+ limits = c(0, max(data$x))) +
+ scale_y_continuous(paste("Cumulative number of relays used for",
+ "introduction points\n")) +
+ scale_colour_hue("Service") +
+ ggtitle(paste("Number of introduction points established on\nthe same",
+ "relay (in the measurement period)\n"))
+
+data <- data[as.character(data$colour) != "agorahooawayyfoe", ]
+ggplot(data) +
+ geom_step(aes(x = x, y = y, colour = as.factor(colour))) +
+ scale_x_continuous("\nIntroduction points established on same relay",
+ limits = c(0, max(data$x, na.rm = TRUE))) +
+ scale_y_continuous(paste("Cumulative number of relays used for",
+ "introduction points\n")) +
+ scale_colour_hue("Service") +
+ ggtitle(paste("Number of introduction points established on\nthe same",
+ "relay (in the measurement period)\n"))
+
diff --git a/task-15513/src/ParseDescriptors.java b/task-15513/src/ParseDescriptors.java
new file mode 100644
index 0000000..ef4d77f
--- /dev/null
+++ b/task-15513/src/ParseDescriptors.java
@@ -0,0 +1,420 @@
+import java.io.BufferedReader;
+import java.io.BufferedWriter;
+import java.io.File;
+import java.io.FileReader;
+import java.io.FileWriter;
+import java.io.IOException;
+import java.io.StringReader;
+import java.text.DateFormat;
+import java.text.ParseException;
+import java.text.SimpleDateFormat;
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+import java.util.SortedMap;
+import java.util.TimeZone;
+import java.util.TreeMap;
+
+import javax.xml.bind.DatatypeConverter;
+
+import org.apache.commons.codec.binary.Base32;
+import org.apache.commons.codec.digest.DigestUtils;
+
+public class ParseDescriptors {
+ public static void main(String[] args) throws IOException,
+ ParseException {
+ /* Parse all hidden-service descriptors in descriptors/ and extract
+ * permanent keys, publication times, and relay identities of
+ * introduction points. */
+ SortedMap<String, /* <- service name */
+ SortedMap<Long, /* <- publication time */
+ List< /* <- list of descriptors */
+ Set<String>>>> /* <- fingerprints and client keys */
+ parsedDescriptors = parseDescriptors(new File("descriptors/"));
+
+ /* Print verbose log. */
+ printVerboseLog(parsedDescriptors, new File("lifetimes-verbose.txt"));
+
+ /* Calculate lifetimes of introduction points, but exclude those that
+ * were contained in the first or last known descriptor of a
+ * service. */
+ writeLifetimes(parsedDescriptors, new File("lifetimes.csv"),
+ "service,lifetime_millis");
+
+ /* Calculate the number of published descriptors per hour, including
+ * descriptor replicas. */
+ writePublishedDescriptors(parsedDescriptors,
+ new File("published-descs.csv"), "service,publication,descs");
+
+ /* Calculate the number of distinct established introduction points
+ * per hour. */
+ writeEstablishedIntroductionPoints(parsedDescriptors,
+ new File("established-intros.csv"), "service,publication,intros");
+
+ writeIntroPointsPerDescriptor(parsedDescriptors,
+ new File("intros-per-desc.csv"),
+ "service,publication,introsperdesc");
+
+ writeIntroductionPointsPerRelay(parsedDescriptors,
+ new File("intros-per-relay.csv"), "service,intros");
+ }
+
+ private static SortedMap<String, SortedMap<Long, List<Set<String>>>>
+ parseDescriptors(File descriptorsDirectory) throws IOException,
+ ParseException {
+ SortedMap<String, SortedMap<Long, List<Set<String>>>>
+ parsedDescriptors = new TreeMap<String, SortedMap<Long,
+ List<Set<String>>>>();
+ for (File descriptorFile : descriptorsDirectory.listFiles()) {
+ String permanentKeyDigest = null;
+ long publicationMillis = -1L;
+ Set<String> introductionPoints = null;
+ BufferedReader br = new BufferedReader(new FileReader(
+ descriptorFile));
+ String line;
+ DateFormat dateTimeFormat = new SimpleDateFormat(
+ "yyyy-MM-dd HH:mm:ss");
+ dateTimeFormat.setLenient(false);
+ dateTimeFormat.setTimeZone(TimeZone.getTimeZone("UTC"));
+ while ((line = br.readLine()) != null) {
+ if (line.equals("permanent-key")) {
+ String permanentKey = appendLines(br,
+ "-----BEGIN RSA PUBLIC KEY-----",
+ "-----END RSA PUBLIC KEY-----");
+ permanentKeyDigest = toDigest(permanentKey);
+ } else if (line.startsWith("publication-time ")) {
+ publicationMillis = dateTimeFormat.parse(line.substring(
+ "publication-time ".length())).getTime();
+ } else if (line.equals("introduction-points")) {
+ String encodedIntroductionPoints = appendLines(br,
+ "-----BEGIN MESSAGE-----", "-----END MESSAGE-----");
+ introductionPoints = parseEncodedIntroductionPoints(
+ encodedIntroductionPoints);
+ }
+ }
+ br.close();
+ if (permanentKeyDigest != null && publicationMillis >= 0L &&
+ introductionPoints != null) {
+ if (!parsedDescriptors.containsKey(permanentKeyDigest)) {
+ parsedDescriptors.put(permanentKeyDigest,
+ new TreeMap<Long, List<Set<String>>>());
+ }
+ if (!parsedDescriptors.get(permanentKeyDigest).containsKey(
+ publicationMillis)) {
+ parsedDescriptors.get(permanentKeyDigest).put(publicationMillis,
+ new ArrayList<Set<String>>());
+ }
+ parsedDescriptors.get(permanentKeyDigest).get(
+ publicationMillis).add(introductionPoints);
+ }
+ }
+ return parsedDescriptors;
+ }
+
+ private static String toDigest(String encodedString) {
+ byte[] bytes = DatatypeConverter.parseBase64Binary(encodedString);
+ String str = new Base32().encodeAsString(DigestUtils.sha(bytes));
+ return str.substring(0, 16).toLowerCase();
+ }
+
+ private static String appendLines(BufferedReader br, String skipLine,
+ String untilLine) throws IOException {
+ StringBuilder appendedLinesBuilder = new StringBuilder();
+ String line = br.readLine();
+ while ((line = br.readLine()) != null) {
+ if (line.equals(skipLine)) {
+ continue;
+ }
+ if (line.equals(untilLine)) {
+ break;
+ }
+ appendedLinesBuilder.append(line);
+ }
+ return appendedLinesBuilder.toString();
+ }
+
+ private static Set<String> parseEncodedIntroductionPoints(
+ String encodedIntroductionPoints) throws IOException {
+ Set<String> introductionPoints = new HashSet<String>();
+ String decodedIntroductionPoints = new String(
+ DatatypeConverter.parseBase64Binary(
+ encodedIntroductionPoints));
+ BufferedReader br = new BufferedReader(new StringReader(
+ decodedIntroductionPoints));
+ String line = null, lastIntroductionPoint = null;
+ while ((line = br.readLine()) != null) {
+ if (line.startsWith("introduction-point ")) {
+ lastIntroductionPoint = line.substring(
+ "introduction-point ".length());
+ } else if (line.equals("service-key")) {
+ String encodedServiceKey = appendLines(br,
+ "-----BEGIN RSA PUBLIC KEY-----",
+ "-----END RSA PUBLIC KEY-----");
+ String serviceKeyDigest = toDigest(encodedServiceKey);
+ introductionPoints.add(lastIntroductionPoint + "-"
+ + serviceKeyDigest);
+ }
+ }
+ br.close();
+ return introductionPoints;
+ }
+
+ private static void printVerboseLog(
+ SortedMap<String, SortedMap<Long, List<Set<String>>>>
+ parsedDescriptors, File verboseLogFile) throws IOException {
+ BufferedWriter bw = new BufferedWriter(new FileWriter(
+ verboseLogFile));
+ DateFormat dateTimeFormat = new SimpleDateFormat(
+ "yyyy-MM-dd HH:mm:ss");
+ dateTimeFormat.setLenient(false);
+ dateTimeFormat.setTimeZone(TimeZone.getTimeZone("UTC"));
+ for (Map.Entry<String, SortedMap<Long, List<Set<String>>>> e0 :
+ parsedDescriptors.entrySet()) {
+ bw.write("\nService " + e0.getKey() + "\n");
+ for (Map.Entry<Long, List<Set<String>>> e1 :
+ e0.getValue().entrySet()) {
+ bw.write(" publication-time "
+ + dateTimeFormat.format(e1.getKey()) + "\n");
+ int descriptor = 0;
+ for (Set<String> intro : e1.getValue()) {
+ bw.write(" descriptor " + descriptor++ + "\n");
+ for (String i : intro) {
+ bw.write(" " + i + "\n");
+ }
+ }
+ }
+ }
+ bw.close();
+ }
+
+ private static void writeLifetimes(
+ SortedMap<String, SortedMap<Long, List<Set<String>>>>
+ parsedDescriptors, File lifetimesCsvFile, String header)
+ throws IOException {
+ SortedMap<String, /* <- service name */
+ List<Long>> /* <- lifetimes in milliseconds */
+ calculatedLifetimes = new TreeMap<String, List<Long>>();
+ for (Map.Entry<String, SortedMap<Long, List<Set<String>>>> e0 :
+ parsedDescriptors.entrySet()) {
+ String permanentKey = e0.getKey();
+ List<Long> lifetimes = new ArrayList<Long>();
+ SortedMap<Long, List<Set<String>>>
+ publicationMillisIntroductionPoints = e0.getValue();
+ long firstPublicationMillis =
+ publicationMillisIntroductionPoints.firstKey();
+ Set<String> introductionPointsInFirstDescriptor =
+ new HashSet<String>();
+ for (Set<String> introductionPoints :
+ publicationMillisIntroductionPoints.get(
+ firstPublicationMillis)) {
+ introductionPointsInFirstDescriptor.addAll(introductionPoints);
+ }
+ Map<String, long[]> liveIntroductionPoints =
+ new HashMap<String, long[]>();
+ for (Map.Entry<Long, List<Set<String>>> e1 :
+ publicationMillisIntroductionPoints.tailMap(
+ firstPublicationMillis + 1L).entrySet()) {
+ long publicationMillis = e1.getKey();
+ Set<String> introductionPoints = new HashSet<String>();
+ for (Set<String> points : e1.getValue()) {
+ introductionPoints.addAll(points);
+ }
+ introductionPointsInFirstDescriptor.retainAll(introductionPoints);
+ for (String introductionPoint : introductionPoints) {
+ if (introductionPointsInFirstDescriptor.contains(
+ introductionPoint)) {
+ /* Skip introduction point, because it was contained in the
+ * first known descriptor of this service, and it could have
+ * been running for a while. */
+ continue;
+ } else if (!liveIntroductionPoints.containsKey(
+ introductionPoint)) {
+ /* This introduction point is new, let's remember it. */
+ liveIntroductionPoints.put(introductionPoint,
+ new long[] { publicationMillis, publicationMillis } );
+ } else {
+ /* This introduction point is not new, and it's still there,
+ * update its last-seen timestamp. */
+ liveIntroductionPoints.get(introductionPoint)[1] =
+ publicationMillis;
+ }
+ }
+ Set<String> deadIntroductionPoints = new HashSet<String>();
+ for (Map.Entry<String, long[]> e2 :
+ liveIntroductionPoints.entrySet()) {
+ String introductionPoint = e2.getKey();
+
+ if (!introductionPoints.contains(introductionPoint)) {
+ long lifetime = e2.getValue()[1] - e2.getValue()[0]
+ + 3600000L;
+ lifetimes.add(lifetime);
+ deadIntroductionPoints.add(introductionPoint);
+ }
+ }
+ for (String introductionPoint : deadIntroductionPoints) {
+ liveIntroductionPoints.remove(introductionPoint);
+ }
+ }
+ calculatedLifetimes.put(permanentKey, lifetimes);
+ }
+ BufferedWriter bw = new BufferedWriter(new FileWriter(
+ lifetimesCsvFile));
+ bw.write(header + "\n");
+ for (Map.Entry<String, List<Long>> e :
+ calculatedLifetimes.entrySet()) {
+ for (long lifetime : e.getValue()) {
+ bw.write(e.getKey() + "," + lifetime + "\n");
+ }
+ }
+ bw.close();
+ }
+
+ private static void writePublishedDescriptors(SortedMap<String,
+ SortedMap<Long, List<Set<String>>>> parsedDescriptors, File csvFile,
+ String header) throws IOException {
+ SortedMap<String, /* <- service name */
+ SortedMap<Long, /* <- publication time */
+ Integer>> /* <- number of published descriptors*/
+ publishedDescriptors =
+ new TreeMap<String, SortedMap<Long, Integer>>();
+ for (Map.Entry<String, SortedMap<Long, List<Set<String>>>> e0 :
+ parsedDescriptors.entrySet()) {
+ String serviceName = e0.getKey();
+ publishedDescriptors.put(serviceName, new TreeMap<Long, Integer>());
+ for (Map.Entry<Long, List<Set<String>>> e1 :
+ e0.getValue().entrySet()) {
+ long publicationMillis = e1.getKey();
+ int descriptors = e1.getValue().size();
+ publishedDescriptors.get(serviceName).put(publicationMillis,
+ descriptors);
+ }
+ }
+ BufferedWriter bw = new BufferedWriter(new FileWriter(csvFile));
+ bw.write(header + "\n");
+ for (Map.Entry<String, SortedMap<Long, Integer>> e0 :
+ publishedDescriptors.entrySet()) {
+ for (Map.Entry<Long, Integer> e1 : e0.getValue().entrySet()) {
+ bw.write(e0.getKey() + "," + e1.getKey() + "," + e1.getValue()
+ + "\n");
+ }
+ }
+ bw.close();
+ }
+
+ private static void writeEstablishedIntroductionPoints(SortedMap<String,
+ SortedMap<Long, List<Set<String>>>> parsedDescriptors, File csvFile,
+ String header) throws IOException {
+ SortedMap<String, /* <- service name */
+ SortedMap<Long, /* <- publication time */
+ Integer>> /* <- number of published descriptors*/
+ establishedIntros =
+ new TreeMap<String, SortedMap<Long, Integer>>();
+ for (Map.Entry<String, SortedMap<Long, List<Set<String>>>> e0 :
+ parsedDescriptors.entrySet()) {
+ String serviceName = e0.getKey();
+ establishedIntros.put(serviceName, new TreeMap<Long, Integer>());
+ for (Map.Entry<Long, List<Set<String>>> e1 :
+ e0.getValue().entrySet()) {
+ long publicationMillis = e1.getKey();
+ Set<String> introductionPoints = new HashSet<String>();
+ for (Set<String> points : e1.getValue()) {
+ introductionPoints.addAll(points);
+ }
+ establishedIntros.get(serviceName).put(publicationMillis,
+ introductionPoints.size());
+ }
+ }
+ BufferedWriter bw = new BufferedWriter(new FileWriter(csvFile));
+ bw.write(header + "\n");
+ for (Map.Entry<String, SortedMap<Long, Integer>> e0 :
+ establishedIntros.entrySet()) {
+ for (Map.Entry<Long, Integer> e1 : e0.getValue().entrySet()) {
+ bw.write(e0.getKey() + "," + e1.getKey() + "," + e1.getValue()
+ + "\n");
+ }
+ }
+ bw.close();
+ }
+
+ private static void writeIntroPointsPerDescriptor(SortedMap<String,
+ SortedMap<Long, List<Set<String>>>> parsedDescriptors,
+ File csvFile, String header) throws IOException {
+ SortedMap<String, SortedMap<Long, List<Integer>>>
+ introPointsPerDescriptor =
+ new TreeMap<String, SortedMap<Long, List<Integer>>>();
+ for (Map.Entry<String, SortedMap<Long, List<Set<String>>>> e0 :
+ parsedDescriptors.entrySet()) {
+ String serviceName = e0.getKey();
+ introPointsPerDescriptor.put(serviceName,
+ new TreeMap<Long, List<Integer>>());
+ for (Map.Entry<Long, List<Set<String>>> e1 :
+ e0.getValue().entrySet()) {
+ long publicationMillis = e1.getKey();
+ introPointsPerDescriptor.get(serviceName).put(publicationMillis,
+ new ArrayList<Integer>());
+ for (Set<String> intros : e1.getValue()) {
+ introPointsPerDescriptor.get(serviceName).get(
+ publicationMillis).add(intros.size());
+ }
+ }
+ }
+ BufferedWriter bw = new BufferedWriter(new FileWriter(csvFile));
+ bw.write(header + "\n");
+ for (Map.Entry<String, SortedMap<Long, List<Integer>>> e0 :
+ introPointsPerDescriptor.entrySet()) {
+ for (Map.Entry<Long, List<Integer>> e1 : e0.getValue().entrySet()) {
+ for (int i : e1.getValue()) {
+ bw.write(e0.getKey() + "," + e1.getKey() + "," + i + "\n");
+ }
+ }
+ }
+ bw.close();
+
+ }
+
+ private static void writeIntroductionPointsPerRelay(SortedMap<String,
+ SortedMap<Long, List<Set<String>>>> parsedDescriptors,
+ File csvFile, String header) throws IOException {
+ SortedMap<String, /* <- service name */
+ List<Integer>> /* <- established introduction point per relay */
+ introductionPointsPerRelay =
+ new TreeMap<String, List<Integer>>();
+ for (Map.Entry<String, SortedMap<Long, List<Set<String>>>> e0 :
+ parsedDescriptors.entrySet()) {
+ String serviceName = e0.getKey();
+ SortedMap<String, Set<String>> clientKeysPerFingerprint =
+ new TreeMap<String, Set<String>>();
+ for (List<Set<String>> descriptors : e0.getValue().values()) {
+ for (Set<String> descriptor : descriptors) {
+ for (String introductionPoint : descriptor) {
+ String[] parts = introductionPoint.split("-");
+ String fingerprint = parts[0];
+ String clientKey = parts[1];
+ if (!clientKeysPerFingerprint.containsKey(fingerprint)) {
+ clientKeysPerFingerprint.put(fingerprint,
+ new HashSet<String>());
+ }
+ clientKeysPerFingerprint.get(fingerprint).add(clientKey);
+ }
+ }
+ }
+ List<Integer> counted = new ArrayList<Integer>();
+ for (Set<String> e1 : clientKeysPerFingerprint.values()) {
+ counted.add(e1.size());
+ }
+ introductionPointsPerRelay.put(serviceName, counted);
+ }
+ BufferedWriter bw = new BufferedWriter(new FileWriter(csvFile));
+ bw.write(header + "\n");
+ for (Map.Entry<String, List<Integer>> e0 :
+ introductionPointsPerRelay.entrySet()) {
+ for (int e1 : e0.getValue()) {
+ bw.write(e0.getKey() + "," + e1 + "\n");
+ }
+ }
+ bw.close();
+ }
+}
More information about the tor-commits
mailing list