the listener thread didn't exit

[David Hill <dd...@zk3.dec.com>]

Hi all,
    I am seeing log messages saying "the listener thread didn't exit". This
is coming out of the housekeeping stuff in the worker module. I am using
2.0.40 and Tru64. I *think* it is happening when the server is trying to
reduce surplus clients.

    On a possbilly related note, I am seeing this when whacking the server
with ab (repeatedly), and after a while seeing the server stop responding
(although in one case it seemed to come back). The server is still acception
connections, just not processing them (at least a telnet connects, but I get
no response)

    Is anyone aware of or chasing this or something like it ?

thanks,
    Dave Hill

Re: the listener thread didn't exit

[David Hill <dd...@zk3.dec.com>]

> This part, though, sounds like a new problem.  Do you have a way to grab
> a stack trace of one of the stuck processes when this happens?

Apparently this is somewhat reproducable, it happened again. Problem is ....
too much to look at, 14 processes, 25 threads per and me hunting for a
needle :-) . Guess I will have to try to tighten down on the runtime
parameters and see if it will fail again.

However, rooting through one of the 14 processes, I seen a trend in the
traces though, many of them are stuck under alloca() under apr_poll. looks
like I will need to talk with my libc folks and see if there are any known
issues under there as I am running on a prerelease version of the OS. It
feels like a locking issue inside alloca() at the moment.

Dave

Re: the listener thread didn't exit

[Brian Pane <br...@apache.org>]

David Hill wrote:

>Hi all,
>    I am seeing log messages saying "the listener thread didn't exit". This
>is coming out of the housekeeping stuff in the worker module. I am using
>2.0.40 and Tru64. I *think* it is happening when the server is trying to
>reduce surplus clients.
>

The worker MPM has known problems with process shutdown,
so this part isn't entirely surprising.

>    On a possbilly related note, I am seeing this when whacking the server
>with ab (repeatedly), and after a while seeing the server stop responding
>(although in one case it seemed to come back). The server is still acception
>connections, just not processing them (at least a telnet connects, but I get
>no response)
>

This part, though, sounds like a new problem.  Do you have a way to grab
a stack trace of one of the stuck processes when this happens?

Thanks,
Brian