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 40afa2b8eb2d90cf9a89e4f7a956bc610191c26d..333dbed55b4409da28081d24a92c22debad49e1f 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
 


@@ -30,9 +31,9 @@ BLACKLIST = ''
 UA_BLOCKLIST = ()
 
 # time before a automatically blocked IP is allowed back
 UA_BLOCKLIST = ()
 
 # time before a automatically blocked IP is allowed back

-BLOCKCOOLOFF_PERIOD=timedelta(hours=1)
+BLOCKCOOLOFF_DELTA=timedelta(hours=1)

 # quiet time before an IP is released from the bulk pool
 # quiet time before an IP is released from the bulk pool

-BULKCOOLOFF_PERIOD=timedelta(minutes=15)
+BULKCOOLOFF_DELTA=timedelta(minutes=15)

 
 BULKLONG_LIMIT=8000
 BULKSHORT_LIMIT=2000
 
 BULKLONG_LIMIT=8000
 BULKSHORT_LIMIT=2000


@@ -50,18 +51,22 @@ MAX_BULK_IPS=85
 try:
     with open(BASEDIR + "/settings/ip_blocks.conf") as f:
         code = compile(f.read(), BASEDIR + "/settings/ip_blocks.conf", 'exec')
 try:
     with open(BASEDIR + "/settings/ip_blocks.conf") as f:
         code = compile(f.read(), BASEDIR + "/settings/ip_blocks.conf", 'exec')

-        exec(code, global_vars, local_vars)
+        exec(code)

 except IOError:
     pass
 
 BLOCK_LIMIT = BLOCK_LOWER
 
 except IOError:
     pass
 
 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'

 
 

-format_pat= re.compile('(?P<ip>[(\d\.)]+) - - \[(?P<time>.*?)\] "(?P<query>.*?)" (?P<return>\d+) (?P<bytes>\d+) "(?P<referer>.*?)" "(?P<ua>.*?)"')
-time_pat= re.compile('[(\d\.)]+ - - \[(?P<time>[^\]]*?)\] ')
+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"
 

+MONTHS = { 'Jan' : 1, 'Feb' : 2, 'Mar' : 3, 'Apr' : 4, 'May' : 5, 'Jun' : 6,
+           'Jul' : 7, 'Aug' : 8, 'Sep' : 9, 'Oct' : 10, 'Nov' : 11, 'Dec' : 12 }
+

 class LogEntry:
     def __init__(self, logline):
         e = format_pat.match(logline)
 class LogEntry:
     def __init__(self, logline):
         e = format_pat.match(logline)


@@ -69,7 +74,8 @@ class LogEntry:
             raise ValueError("Invalid log line:", logline)
         e = e.groupdict()
         self.ip = e['ip']
             raise ValueError("Invalid log line:", logline)
         e = e.groupdict()
         self.ip = e['ip']

-        self.date = datetime.strptime(e['time'], logtime_pat).replace(tzinfo=None)
+        self.date = datetime(int(e['t_year']), MONTHS[e['t_month']], int(e['t_day']),
+                             int(e['t_hour']), int(e['t_min']), int(e['t_sec']))

         qp = e['query'].split(' ', 2) 
         if len(qp) < 2:
             self.request = None
         qp = e['query'].split(' ', 2) 
         if len(qp) < 2:
             self.request = None


@@ -79,11 +85,11 @@ class LogEntry:
             if qp[0] == 'OPTIONS':
                 self.request = None
             else:
             if qp[0] == 'OPTIONS':
                 self.request = None
             else:

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

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

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

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

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

                     self.request = 'D'
                 else:
                     self.request = None
                     self.request = 'D'
                 else:
                     self.request = None


@@ -97,7 +103,9 @@ class LogEntry:
         if e is None:
             return None
         e = e.groupdict()
         if e is None:
             return None
         e = e.groupdict()

-        return datetime.strptime(e['time'], logtime_pat).replace(tzinfo=None)
+        #return datetime.strptime(e['time'], logtime_pat).replace(tzinfo=None)
+        return datetime(int(e['t_year']), MONTHS[e['t_month']], int(e['t_day']),
+                             int(e['t_hour']), int(e['t_min']), int(e['t_sec']))

 
 
 class LogFile:
 
 
 class LogFile:


@@ -117,28 +125,72 @@ class LogFile:
         return LogEntry.get_log_time(l) if l is not None else None
 
     def seek_to_date(self, target):
         return LogEntry.get_log_time(l) if l is not None else None
 
     def seek_to_date(self, target):

-        date1 = self.seek_next(0)
-        if date1 > target:
+        # start position for binary search
+        fromseek = 0
+        fromdate = self.seek_next(0)
+        if fromdate > target:
+            return True
+        # end position for binary search
+        toseek = -100
+        while -toseek < self.len:
+            todate = self.seek_next(self.len + toseek)
+            if todate is not None:
+                break
+            toseek -= 100
+        if todate is None or todate < target:

             return False
             return False

-        curseek = 2*1024*1024
-        curdate = self.seek_next(curseek)
-        if curdate is None:
-            raise RuntimeError("Cannot seek to %d" % curseek)
-        while abs((target - curdate).total_seconds()) > 1.0:
-            bps = curseek / ((curdate - date1).total_seconds())
-            curseek += (target - curdate).total_seconds() * bps
-            if curseek < 0:
-                curseek = 0
-            elif curseek > self.len:
-                curseek = self.len - bps
-            curdate = self.seek_next(curseek)
-            if curdate is None:
-                raise RuntimeError("Cannot see to %d" % curseek)
-        return True
+        toseek = self.len + toseek
+
+
+        while True:
+            bps = (toseek - fromseek) / (todate - fromdate).total_seconds()
+            newseek = fromseek + int((target - fromdate).total_seconds() * bps)
+            newdate = self.seek_next(newseek)
+            if newdate is None:
+                return False;
+            error = abs((target - newdate).total_seconds())
+            if error < 1:
+                return True
+            if newdate > target:
+                toseek = newseek
+                todate = newdate
+                oldfromseek = fromseek
+                fromseek = toseek - error * bps
+                while True:
+                    if fromseek <= oldfromseek:
+                        fromseek = oldfromseek
+                        fromdate = self.seek_next(fromseek)
+                        break
+                    fromdate = self.seek_next(fromseek)
+                    if fromdate < target:
+                        break;
+                    bps *=2
+                    fromseek -= error * bps
+            else:
+                fromseek = newseek
+                fromdate = newdate
+                oldtoseek = toseek
+                toseek = fromseek + error * bps
+                while True:
+                    if toseek > oldtoseek:
+                        toseek = oldtoseek
+                        todate = self.seek_next(toseek)
+                        break
+                    todate = self.seek_next(toseek)
+                    if todate > target:
+                        break
+                    bps *=2
+                    toseek += error * bps
+            if toseek - fromseek < 500:
+                return True
+

 
     def loglines(self):
         for l in self.fd:
 
     def loglines(self):
         for l in self.fd:

-            yield LogEntry(l)
+            try:
+                yield LogEntry(l)
+            except ValueError:
+                pass # ignore invalid lines

 
 class BlockList:
 
 
 class BlockList:
 


@@ -154,9 +206,9 @@ class BlockList:
                 ip, typ = line.strip().split(' ')
                 if ip not in self.blacklist:
                     if typ == 'block':
                 ip, typ = line.strip().split(' ')
                 if ip not in self.blacklist:
                     if typ == 'block':

-                        prevblocks.add(ip)
+                        self.prevblocks.add(ip)

                     elif typ == 'bulk':
                     elif typ == 'bulk':

-                        prevbulks.add(ip)
+                        self.prevbulks.add(ip)

             fd.close()
         except IOError:
             pass #ignore non-existing file
             fd.close()
         except IOError:
             pass #ignore non-existing file


@@ -217,7 +269,7 @@ if __name__ == '__main__':
         sys.exit(-1)
 
     if len(sys.argv) == 2:
         sys.exit(-1)
 
     if len(sys.argv) == 2:

-        dt = datetime.now() - BLOCKCOOLOFF_PERIOD
+        dt = datetime.now() - BLOCKCOOLOFF_DELTA

     else:
         dt = datetime.strptime(sys.argv[2], "%Y-%m-%d %H:%M:%S")
 
     else:
         dt = datetime.strptime(sys.argv[2], "%Y-%m-%d %H:%M:%S")
 


@@ -230,7 +282,7 @@ if __name__ == '__main__':
 
     bl = BlockList()
 
 
     bl = BlockList()
 

-    shortstart = dt + BLOCKCOOLOFF_PERIOD - BULKCOOLOFF_PERIOD
+    shortstart = dt + BLOCKCOOLOFF_DELTA - BULKCOOLOFF_DELTA

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


@@ -252,11 +304,14 @@ if __name__ == '__main__':
     fd = open("/proc/loadavg")
     cpuload = int(float(fd.readline().split()[2]))
     fd.close()
     fd = open("/proc/loadavg")
     cpuload = int(float(fd.readline().split()[2]))
     fd.close()

-    dbload = total200 / BULKCOOLOFF_PERIOD.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
     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


@@ -298,7 +353,7 @@ 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)