I've submitted a patch which appears to resolve this issue in my setup. Josh On 02/19/2013 11:31 AM, Josh Hunt wrote:
I just ran into this and figured I'd see if anyone else has hit the issue. If I create a set with timeouts enabled and then add entries to this set which requires a resize, the timeout values become corrupt. Here's an example of what I'm doing: ipset create foo hash:ip,port,ip maxelem 16536 timeout 600 Then I add ~16k entries to the set and I see things like this as I'm adding entries to the set: Name: foo Type: hash:ip,port,ip Revision: 1 Header: family inet hashsize 2048 maxelem 16536 timeout 600 Size in memory: 268456 References: 0 Members: 2.0.3.233,tcp:9999,172.24.196.41 timeout 1610614 2.0.17.81,tcp:9999,172.24.196.41 timeout 596 2.0.21.190,tcp:9999,172.24.196.41 timeout 598 2.0.4.156,tcp:9999,172.24.196.41 timeout 1610879 2.0.6.132,tcp:9999,172.24.196.41 timeout 1611592 I believe this is resizing related b/c if I force the hashsize to something that won't get resized (ex: something >= maxelem) then I don't see this problem. I noticed in the resize code there's a lot of casting from type_pf_elem to type_pf_telem, but haven't investigated yet to determine the root cause of the problem. I've reproduced this on kernel 3.2 with the latest ipset userspace utils 6.16.1. Thanks for any help Josh
-- To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
