Hi Daniel,

you are absolutely right and that part of code is for sure wrong. Those lines have been there since the beginning and for sure due to my lack of understanding of transactions's internals and avps (don't remember why sincerely).

For the records, I've been using mixed suspended/not suspended from the beginning but I've always used separate return routes for suspended and not suspended queries. That could explain why I never had it.

Thanks for looking into it

On Sat, Apr 18, 2020 at 3:13 AM Alex Balashov <notifications@github.com> wrote:

@miconda To answer a few other questions:

1. I always set $http_req(suspend) = 1 on async HTTP client transactions;

2. There are no event_routes anywhere;

3. The flow is complex, though. After this async HTTP query, there is another synchronous HTTP query using the normal http_client module + mqueue + rtimer -- it was implemented this way because of some problems with transaction-persistent variables (ironically) when suspending the same TM transaction twice, and a desire for more granular control over the request pipelines given the high and somewhat unpredictable latency of what's being queried.

Nevertheless, all this takes place long after the crash point.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or unsubscribe.

_______________________________________________
Kamailio (SER) - Development Mailing List
sr-dev@lists.kamailio.org
https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-dev