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 1858266734bc18d6f20cf91c70321596414909bb..97bad8da118e96b1f4287873919287389a3e2cd8 100755 (executable)
--- a/utils/cron_ipanalyse.py
+++ b/utils/cron_ipanalyse.py
@@ -9,6 +9,7 @@
 import re
 import os
 import sys
 import re
 import os
 import sys

+import subprocess

 from datetime import datetime, timedelta
 from collections import defaultdict
 
 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)
 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
 
 BULKLONG_LIMIT=8000
 BULKSHORT_LIMIT=2000


@@ -58,8 +61,8 @@ BLOCK_LIMIT = BLOCK_LOWER
 
 time_regex = r'(?P<t_day>\d\d)/(?P<t_month>[A-Za-z]+)/(?P<t_year>\d\d\d\d):(?P<t_hour>\d\d):(?P<t_min>\d\d):(?P<t_sec>\d\d) [+-]\d\d\d\d'
 
 
 time_regex = r'(?P<t_day>\d\d)/(?P<t_month>[A-Za-z]+)/(?P<t_year>\d\d\d\d):(?P<t_hour>\d\d):(?P<t_min>\d\d):(?P<t_sec>\d\d) [+-]\d\d\d\d'
 

-format_pat= re.compile(r'(?P<ip>[(\d\.)]+) - - \['+ time_regex + r'] "(?P<query>.*?)" (?P<return>\d+) (?P<bytes>\d+) "(?P<referer>.*?)" "(?P<ua>.*?)"')
-time_pat= re.compile(r'[(\d\.)]+ - - \[' + time_regex + '\] ')
+format_pat= re.compile(r'(?P<ip>[a-f\d\.:]+) - - \['+ time_regex + r'] "(?P<query>.*?)" (?P<return>\d+) (?P<bytes>\d+) "(?P<referer>.*?)" "(?P<ua>.*?)"')
+time_pat= re.compile(r'[a-f\d:\.]+ - - \[' + time_regex + '\] ')

 
 logtime_pat = "%d/%b/%Y:%H:%M:%S %z"
 
 
 logtime_pat = "%d/%b/%Y:%H:%M:%S %z"
 


@@ -84,12 +87,16 @@ class LogEntry:
             if qp[0] == 'OPTIONS':
                 self.request = None
             else:
             if qp[0] == 'OPTIONS':
                 self.request = None
             else:

-                if qp[1].startswith('/search'):
+                if '/?' in qp[1]:

                     self.request = 'S'
                     self.request = 'S'

-                elif qp[1].startswith('/reverse'):
+                elif '/search' in qp[1]:
+                    self.request = 'S'
+                elif '/reverse' in qp[1]:

                     self.request = 'R'
                     self.request = 'R'

-                elif qp[1].startswith('/details'):
+                elif '/details' in qp[1]:

                     self.request = 'D'
                     self.request = 'D'

+                elif '/lookup' in qp[1]:
+                    self.request = 'L'

                 else:
                     self.request = None
         self.query = e['query']
                 else:
                     self.request = None
         self.query = e['query']


@@ -216,14 +223,18 @@ class BlockList:
 class IPstats:
 
     def __init__(self):
 class IPstats:
 
     def __init__(self):

+        self.redirected = 0

         self.short_total = 0
         self.short_api = 0
         self.long_total = 0
         self.long_api = 0
         self.short_total = 0
         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):
         self.long_total += 1
         self.bad_ua = False
 
     def add_long(self, logentry):
         self.long_total += 1

+        if logentry.retcode == 301:
+            return

         if logentry.request is not None:
             self.long_api += 1
         if not self.bad_ua:
         if logentry.request is not None:
             self.long_api += 1
         if not self.bad_ua:


@@ -232,20 +243,31 @@ class IPstats:
 
     def add_short(self, logentry):
         self.short_total += 1
 
     def add_short(self, logentry):
         self.short_total += 1

+        if logentry.retcode == 301:
+            self.redirected += 1
+            return

         if logentry.request is not None:
             self.short_api += 1
         self.add_long(logentry)
 
         if logentry.request is not None:
             self.short_api += 1
         self.add_long(logentry)
 

+    def add_block(self, logentry):
+        self.block_total += 1
+
+    def ignores_warnings(self, wasblocked):
+        return self.block_total > 5 or (wasblocked and self.redirected > 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)
     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'
-        if self.long_api > BLOCK_UPPER or self.short_api > BLOCK_UPPER / 3:
+            return None if self.long_total < 20 else 'block'
+        if self.long_api > BLOCK_UPPER \
+            or self.short_api > BLOCK_UPPER / 3 \
+            or (self.redirected > 100 and self.short_total == self.redirected):

                 # client totally overdoing it
                 return 'block'
         if was_bulked:
                 # 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:
                 # client has stopped, debulk
                 return None
             if self.long_api > BLOCK_LIMIT or self.short_api > BLOCK_LIMIT / 3:


@@ -254,8 +276,8 @@ class IPstats:
             return 'bulk'
 
         if self.long_api > BULKLONG_LIMIT or self.short_api > BULKSHORT_LIMIT:
             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
             return 'bulk'
 
         return None


@@ -282,6 +304,7 @@ if __name__ == '__main__':
     bl = BlockList()
 
     shortstart = dt + BLOCKCOOLOFF_DELTA - BULKCOOLOFF_DELTA
     bl = BlockList()
 
     shortstart = dt + BLOCKCOOLOFF_DELTA - BULKCOOLOFF_DELTA

+    blockstart = dt + BLOCKCOOLOFF_DELTA - BLOCKCHECK_DELTA

     notlogged = bl.whitelist | bl.blacklist
 
     stats = defaultdict(IPstats)
     notlogged = bl.whitelist | bl.blacklist
 
     stats = defaultdict(IPstats)


@@ -298,22 +321,28 @@ if __name__ == '__main__':
             stats[l.ip].add_short(l)
         if l.request is not None and l.retcode == 200:
             total200 += 1
             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()
 
     # 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)
 
     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:
     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
 
 
     # collecting statistics


@@ -349,11 +378,11 @@ if __name__ == '__main__':
             elif wasbulked:
                 debulked.append(k)
     for i in bl.blacklist:
             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)
     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)))
     fd = open(LOGFILE, 'a')
     if unblocked:
         fd.write(logstr % ('unblocked:', ', '.join(unblocked)))


@@ -367,4 +396,7 @@ if __name__ == '__main__':
         fd.write(logstr % (' ua block:', ', '.join(uablocked)))
     if blocked:
         fd.write(logstr % ('new block:', ', '.join(blocked)))
         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(k in bl.prevblocks) and k not in notlogged and ':' not in k:
+    #        fd.write(logstr % ('Warning ignored:', k))

     fd.close()
     fd.close()