[stunnel-users] Stunnel 4.21 99% cpu ultilization

[Paul Dekkers]

Hi,

Fat Wallet wrote:

> If anyone have a clue on what I should do to fix this issue, I would
> really appreciated:
> 
> One of the stunnel process is running at 99% and in R state and the time
> just keep incrementing.

[...]

> Fat Wallet wrote:  [Mon Jun 30 2008, 01:28:11PM EDT]
>> Can someone please help with this issue?  This 100% cpu
>> utilization issue happens randomly on multiprocess cpu.
> 
> strace, ltrace, build stunnel with debugging symbols and attach
> gdb to the spinning process, etc.

I'd just like to add that I've seen similar behavior with stunnel 4.20
on FreeBSD (6.2). I upgraded to 4.20 from 4.07 a while ago, and that one
was a lot more stable.

Thing is, if it consumes 99% CPU in our setup, all of the sessions
stall. I'm running with monit for a while now, looking after stunnel and
restarting it when necessary.

Far from ideal. It's hard to reproduce, and I couldn't find a cause with
strace or ktrace / kdump (it actually seems like there's not much going
on when the process stalls), and I didn't have the time to dig deeper.
(Or report it, actually ;-)) And I'm afraid that adding debugging
symbols would harm the performance that we need.

Just every now and then I see someone posting the same issue on the list
(there was one before). So maybe it's time to throw in a "me too". I'm
afraid I can't help a lot though with debugging this issue. It's on a
production server, and I have no clue when it actually happens. It just
does, every now and then, once a month or so.

Regards,
Paul