Hi there,
We're using tinc to create a vpn between two halfs of a linux cluster
situated in geographically disparate locations. As time progresses, the
two tinc daemons use more and more memory, to the point that after a few
days running, their vm size is around the 400 megabyte mark.
The version of tinc that we're using is 1.0.3, and we are running on a
'slightly' customised Gentoo.
I found mention of a memory leak in the mail list archive, but it was in
reference to a bug in the OpenSSL man pages, which has since been fixed.
I've tried rebuilding tinc without --enable-jumbograms, and it appears
to have slown the leak down (although I've only had this new build
running for about 24 hours).
It might be worth mentioning that the increase in memory usage is
constant, and when graphed via rrdtool was very much a straight line.
Has anyone experienced anything similar to this? Is it a known problem?
I can give far more details if they are needed.
Thanks, Scott.
--
Scott Mohekey
Analyst Programmer
scott.mohekey@telogis.com
www.telogis.com
"Dedicated to increasing the value of your fleet."