SSN_START and SSN_CLOSE hooks can be useful for tracking/managing session
level attributes such as context on TCP connection, rate/number of transactions
within the session etc. I'd vote to keep the hooks as it feels very useful even
though there isn't a plugin currently using them.
On Thursday, July 11, 2019, 3:45:31 PM PDT, Walt Karas
<[email protected]> wrote:
But isn't an example using kruft also kruft?
I think the only scenario for this is you hook a continuation globally on
SSN_START. You poke and prod the particular session and decide there is a
reason to sett hooks for each transaction in the session. The alternative
is the have a global hook for TXN_START. Looking at the two examples, the
handling code for SSN_START unconditionally adds the TXN_START hook. So
they seem to be pointlessly added complexity.
traffic_dump, in its SSN_START global hook, does some error checking before
setting a hook for all transactions in the session. But it's not clear
that's a big improvement over doing the checks in a global TXN_START hook.
Seems like this is kruft if we can't think of a clear example where it's
crucial.
On Thu, Jul 11, 2019 at 5:11 PM Leif Hedstrom <[email protected]> wrote:
>
>
> > On Jul 11, 2019, at 3:58 PM, Walt Karas <[email protected]>
> wrote:
> >
> > Should we deprecate this API function?
> >
> > The only core plugin that uses it is traffic_dump (experimental). I
> don't
> > see it used anywhere in VzM internal plugins. Anyone using it in their
> > internal plugins? (Zwoop of course can't say without being disappeared
> to
> > a blacksite, which actually would probably be nice, in a warm climate
> with
> > good beaches and drinks with little umbrellas in them.)
>
> It seems this is somewhere between a global hook and a remap / txn hook?
> I.e. it’s not quite global, but applies to every transaction on this
> session?
>
> That seems potentially useful, but I don’t know of any actual usage
> either. I did notice there’s one example plugin using it as well.
>
> — Leif
>
> >
> > I'm trying to muck around in the CPPAPI again. The existence of this
> call
> > implies a need for a new SessionPlugin type, which I fear would just be
> > unused kruft.
>
>
