Merge remote-tracking branch 'upstream/master'

[nominatim.git] / utils / cron_ipanalyse.py

diff --git a/utils/cron_ipanalyse.py b/utils/cron_ipanalyse.py
index 4dddb1cf35447d8e20612574293843650cf14709..1ca2267fa4144081020db88545538fb01e76559c 100755 (executable)
--- a/utils/cron_ipanalyse.py
+++ b/utils/cron_ipanalyse.py
@@ -9,6 +9,7 @@
 import re
 import os
 import sys
+import subprocess
 from datetime import datetime, timedelta
 from collections import defaultdict
 
@@ -33,6 +34,8 @@ UA_BLOCKLIST = ()
 BLOCKCOOLOFF_DELTA=timedelta(hours=1)
 # quiet time before an IP is released from the bulk pool
 BULKCOOLOFF_DELTA=timedelta(minutes=15)
+# time to check if new accesses appear despite being blocked
+BLOCKCHECK_DELTA=timedelta(minutes=1)
 
 BULKLONG_LIMIT=8000
 BULKSHORT_LIMIT=2000
@@ -84,12 +87,16 @@ class LogEntry:
             if qp[0] == 'OPTIONS':
                 self.request = None
             else:
-                if '/search' in qp[1]:
+                if '/?' in qp[1]:
+                    self.request = 'S'
+                elif '/search' in qp[1]:
                     self.request = 'S'
                 elif '/reverse' in qp[1]:
                     self.request = 'R'
                 elif '/details' in qp[1]:
                     self.request = 'D'
+                elif '/lookup' in qp[1]:
+                    self.request = 'L'
                 else:
                     self.request = None
         self.query = e['query']
@@ -220,6 +227,7 @@ class IPstats:
         self.short_api = 0
         self.long_total = 0
         self.long_api = 0
+        self.block_total = 0
         self.bad_ua = False
 
     def add_long(self, logentry):
@@ -236,16 +244,22 @@ class IPstats:
             self.short_api += 1
         self.add_long(logentry)
 
+    def add_block(self, logentry):
+        self.block_total += 1
+
+    def ignores_warnings(self):
+        return self.block_total > 5
+
     def new_state(self, was_blocked, was_bulked):
         if was_blocked:
             # deblock only if the IP has been really quiet
             # (properly catches the ones that simply ignore the HTTP error)
-            return None if self.long_total < 5 else 'block'
+            return None if self.long_total < 20 else 'block'
         if self.long_api > BLOCK_UPPER or self.short_api > BLOCK_UPPER / 3:
                 # client totally overdoing it
                 return 'block'
         if was_bulked:
-            if self.short_total < 5:
+            if self.short_total < 20:
                 # client has stopped, debulk
                 return None
             if self.long_api > BLOCK_LIMIT or self.short_api > BLOCK_LIMIT / 3:
@@ -254,8 +268,8 @@ class IPstats:
             return 'bulk'
 
         if self.long_api > BULKLONG_LIMIT or self.short_api > BULKSHORT_LIMIT:
-            if self.bad_ua:
-                return 'uablock' # bad useragent
+            #if self.bad_ua:
+            #    return 'uablock' # bad useragent
             return 'bulk'
 
         return None
@@ -282,6 +296,7 @@ if __name__ == '__main__':
     bl = BlockList()
 
     shortstart = dt + BLOCKCOOLOFF_DELTA - BULKCOOLOFF_DELTA
+    blockstart = dt + BLOCKCOOLOFF_DELTA - BLOCKCHECK_DELTA
     notlogged = bl.whitelist | bl.blacklist
 
     stats = defaultdict(IPstats)
@@ -298,22 +313,28 @@ if __name__ == '__main__':
             stats[l.ip].add_short(l)
         if l.request is not None and l.retcode == 200:
             total200 += 1
+        if l.date > blockstart and l.retcode in (403, 429):
+            stats[l.ip].add_block(l)
 
     # adapt limits according to CPU and DB load
     fd = open("/proc/loadavg")
     cpuload = int(float(fd.readline().split()[2]))
     fd.close()
-    dbload = total200 / BULKCOOLOFF_DELTA.total_seconds()
+    # check the number of excess connections to apache
+    dbcons = int(subprocess.check_output("netstat -s | grep 'connections established' | sed 's:^\s*::;s: .*::'", shell=True))
+    fpms = int(subprocess.check_output('ps -Af | grep php-fpm | wc -l', shell=True))
+    dbload = max(0, dbcons - fpms)
 
     numbulks = len(bl.prevbulks)
-    BLOCK_LIMIT = max(BLOCK_LIMIT, BLOCK_UPPER - BLOCK_LOADFAC * (dbload - 75))
-    BULKLONG_LIMIT = max(BULK_LOWER, BULKLONG_LIMIT - BULK_LOADFAC * (cpuload - 14))
+    BLOCK_LIMIT = max(BLOCK_LIMIT, BLOCK_UPPER - BLOCK_LOADFAC * dbload)
+    BULKLONG_LIMIT = max(BULK_LOWER, BULKLONG_LIMIT - BULK_LOADFAC * cpuload)
     if numbulks > MAX_BULK_IPS:
         BLOCK_LIMIT = max(3600, BLOCK_LOWER - (numbulks - MAX_BULK_IPS)*10)
     # if the bulk pool is still empty, clients will be faster, avoid having
     # them blocked in this case
     if numbulks < 10:
-        BLOCK_LIMIT = 2*BLOCK_UPPER
+        BLOCK_UPPER *= 2
+        BLOCK_LIMIT = BLOCK_UPPER
 
 
     # collecting statistics
@@ -349,11 +370,11 @@ if __name__ == '__main__':
             elif wasbulked:
                 debulked.append(k)
     for i in bl.blacklist:
-        fd.write("%s ban\n" % k)
+        fd.write("%s ban\n" % i)
     fd.close()
 
     # TODO write logs (need to collect some statistics)
-    logstr = datetime.now().strftime('%Y-%m-%d %H:%M') + ' %s %s\n'
+    logstr = datetime.now().strftime('%d/%b/%Y:%H:%M:%S') + ' %s %s\n'
     fd = open(LOGFILE, 'a')
     if unblocked:
         fd.write(logstr % ('unblocked:', ', '.join(unblocked)))
@@ -367,4 +388,7 @@ if __name__ == '__main__':
         fd.write(logstr % (' ua block:', ', '.join(uablocked)))
     if blocked:
         fd.write(logstr % ('new block:', ', '.join(blocked)))
+    for k,v in stats.items():
+        if v.ignores_warnings() and k not in notlogged and ':' not in k:
+            fd.write(logstr % ('Warning ignored:', k))
     fd.close()