I'm using Judy to insert IP addresses from web requests
as the Index with a custom Value, along with the time of
insertion for the ip address in a separate JudyL array.
Every 10 seconds or so I iterate through the Judy arrays
to dump the current contents of the judy array to a
hash file on disk, and to expire the judy entries based on
the times of insertion. With just 128 ip addresses
memory usage has already grown to 1548 as reported by
JudyLMemUsed. After a day I see about 1gig used
according to top, with the 10 second snapshot file still
showing ip's in the low hundreds. Is there some
recommended pseudo code to deal with such situations
in a robust manner or common type of race condition I
should be looking out for?
Log in to post a comment.