Merge remote-tracking branch 'upstream/master'

[nominatim.git] / utils / cron_banip.py

diff --git a/utils/cron_banip.py b/utils/cron_banip.py
index d716affbe8e153b99d3fc8810c36ee61dbde9b19..ad32b852f5159fdca5f3f4b7620434abadd5d01b 100755 (executable)
--- a/utils/cron_banip.py
+++ b/utils/cron_banip.py
@@ -61,8 +61,11 @@ BULKCOOLOFF_PERIOD='15 min'
 BULKLONG_LIMIT=8000
 BULKSHORT_LIMIT=2000
 BLOCK_UPPER=19000
-BLOCK_LOADFAC=300
-BULK_LOADFAC=100
+BLOCK_LOWER=4000
+BLOCK_LOADFAC=380
+BULK_LOADFAC=160
+BULK_LOWER=1500
+MAX_BULK_IPS=85
 
 #
 # END OF DEFAULT SETTINGS
@@ -73,11 +76,6 @@ try:
 except IOError:
     pass
 
-# determine current load
-fd = open("/proc/loadavg")
-avgload = int(float(fd.readline().split()[1]))
-fd.close()
-
 # read the previous blocklist
 WHITELIST = set(WHITELIST.split()) if WHITELIST else set()
 prevblocks = []
@@ -99,25 +97,31 @@ try:
 except IOError:
     pass #ignore non-existing file
 
-# current number of bulks
-numbulks = len(prevbulks)
-
-BLOCK_LIMIT = BLOCK_UPPER - BLOCK_LOADFAC * (numbulks - 30)
-BULKLONG_LIMIT = BULKLONG_LIMIT - BULK_LOADFAC * (avgload - 16)
-
+# determine current load
+fd = open("/proc/loadavg")
+avgload = int(float(fd.readline().split()[2]))
+fd.close()
+# DB load
 conn = psycopg2.connect('dbname=nominatim')
 cur = conn.cursor()
+cur.execute("select count(*)/60 from new_query_log where starttime > now() - interval '1min'")
+dbload = int(cur.fetchone()[0])
+
+BLOCK_LIMIT = max(BLOCK_LOWER, BLOCK_UPPER - BLOCK_LOADFAC * (dbload - 75))
+BULKLONG_LIMIT = max(BULK_LOWER, BULKLONG_LIMIT - BULK_LOADFAC * (avgload - 14))
+if len(prevbulks) > MAX_BULK_IPS:
+    BLOCK_LIMIT = max(3600, BLOCK_LOWER - (len(prevbulks) - MAX_BULK_IPS)*10)
 
 # get the new block candidates
 cur.execute("""
   SELECT ipaddress, max(count) FROM
    ((SELECT * FROM
-     (SELECT ipaddress, sum(CASE WHEN type = 'search' THEN 3 ELSE 1 END) as count FROM new_query_log
+     (SELECT ipaddress, sum(case when endtime is null then 1 else 1+1.5*date_part('epoch',endtime-starttime) end) as count FROM new_query_log
       WHERE starttime > now() - interval '1 hour' GROUP BY ipaddress) as i
    WHERE count > %s)
    UNION
-   (SELECT ipaddress, count * 4 FROM
-     (SELECT ipaddress, sum(CASE WHEN type = 'search' THEN 2 ELSE 1 END) as count FROM new_query_log 
+   (SELECT ipaddress, count * 3 FROM
+     (SELECT ipaddress, sum(case when endtime is null then 1 else 1+1.5*date_part('epoch',endtime-starttime) end) as count FROM new_query_log 
       WHERE starttime > now() - interval '10 min' GROUP BY ipaddress) as i
    WHERE count > %s)) as o
   GROUP BY ipaddress
@@ -182,6 +186,7 @@ if debulkcandidates:
     cur.execute("""
         SELECT DISTINCT ipaddress FROM new_query_log
         WHERE ipaddress IN ('%s') AND starttime > now() - interval '%s'
+        AND starttime > date_trunc('day', now())
         """ % ("','".join(debulkcandidates), BULKCOOLOFF_PERIOD))
 
     for c in cur: