vmod_redis

[Jason Price]

Followup question:  How would someone measure the current use of
thread_pool_workspace?

On Sat, Nov 23, 2013 at 8:59 PM, Jason Price <japrice at gmail.com> wrote:
> I'm working with vmod_redis to do some metrics gathering and sundry.
> But I think I've got a memory leak in there.
>
> Is there a recent, canonical implementation of vmod_redis?  I've
> worked my way through github a couple of times trying to find 'the
> one', but to no avail.  (ideally, the vmod_redis implementation would
> support the pipeline call)
>
> My symptoms:  Varnish, vmod_redis, and redis all work beautifully, but
> after several hours, the redis connections start returning "Out of
> memory" errors.  The server has stacks (10+GB) of unused ram, so
> that's not the problem.  I believe my threads are bumping into
> 'thread_pool_workspace" or a similar limit, but I'm having challenges
> proving that assertion.  I know redis isn't the problem because other
> clients can connect to it without issue at the same time varnish
> can't.
>
> Thank you for any pointers;
> Jason