[
http://jira.jboss.com/jira/browse/JGRP-591?page=comments#action_12376843 ]
Bela Ban commented on JGRP-591:
-------------------------------
Got it. Yes, hashCode() has collisions, like any non-perfect hash. I *don't* think
though that String.hashCode() is a perfect hash either.
There is a (small) chance that, if the hash code and port of 2 IpAddresses are the same,
we will have identical objects, leading to incorrect behavior. Again, this chance is very
small.
A solution to this would be to compare the string resulting from IP address and port,
e.g.
"1.2.3.4:2" versus "1.2.3.5:1". This would always work as there cannot
be 2 identical IP addresses which have the same port open.
With logical addresses (
http://jira.jboss.com/jira/browse/JGRP-129), we will create
cluster-wide unique identifiers, so this issue will go away.
IpAddress.hashCode() not unique per ip/port
-------------------------------------------
Key: JGRP-591
URL:
http://jira.jboss.com/jira/browse/JGRP-591
Project: JGroups
Issue Type: Bug
Affects Versions: 2.4.1 SP4
Environment: windows xp sp2, jdk 1.6 update 2
Reporter: robert chou
Assigned To: Bela Ban
Fix For: 2.4.1 SP5, 2.5.1, 2.6
Current implementation produces hashcode where ip/port combos "1.2.3.4:2" and
"1.2.3.5:1" result in the same hash code.
Existing code:
public final int hashCode() {
return ip_addr != null ? ip_addr.hashCode() + port : port;
}
Manual test code:
InetAddress ip1 = InetAddress.getByAddress( new byte[] {1,2,3,4} );
int port1 = 2;
int hash1 = ip1.hashCode() + port1;
System.out.println( "hash1 = " + hash1 );
InetAddress ip2 = InetAddress.getByAddress( new byte[] {1,2,3,5} );
int port2 = 1;
int hash2 = ip2.hashCode() + port2;
System.out.println( "hash2 = " + hash2 );
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira