Hi Cyril, I've just committed and pushed a fix for the double locking that happens when calling pendconn_redistribute() from srv_update_status() and it fixes your report (just checked). Actually I'm realising that I should have credited you and Francis for the report since it's the same one :-(
However be careful, as for me there's still a concurrency issue on
lbprm.set_server_status_{up,down} and set_server_weight(). I'm now
checking how to address this.
Thanks!
Willy
