Colin Keith wrote: > I want to start by saying that I don't know if this is an exim issue or a
> BIND issue. It would seem as though it is a BIND issue because its about
> DNS but I'm curious if anyone else has seen this.
Something sounds catastrophically wrong there. I've tried to reproduce
the issue here rather simplistically but I have a feeling our
environments differ a bit. Every new SMTP connection should be getting
its own separate forked process with its own memory space which means
there is something truly weird going on to end up sharing a PTR record
between multiple processes.
Could you post the Local/Makefile used to compile exim, "ldd -v
/path/to/exim-4.69", your config file options for verify_hosts and maybe
the connect ACL?
Can you try flushing and restarting your resolver while it's in the
middle of this issue too? That should discern between which one has the
issue.
So far I'm thinking of some fairly outlandish ideas for what it might be.
* your ram is fragged
* your resolver is fragged
* maybe a strange bug with the light forking done by linux that does a
copy on first use of memory but it somehow manages to use the parent
memory space the first time and all subsequent forks copy the memory
from the parent and don't clear it.
* resolver library ended up with the same chunk of ram for a
particular process and it doesn't clear the memory before use. Should
eventually be fixed by a good lookup with that memory chunk.
* something not getting cleared somewhere :P
* The static memory used by "gethostbyaddr()" is somehow not getting
cleared and magically returns the same thing all the time?
