We occasionally see that a verto user loses access to a conference for ~15 seconds. Here's an example from our FS v10.5 / Debian-Buster logging. Note the BAD READ which seems to be the cause:---------------------------------------------------------
82133cbd-8d22-4be4-a52a-e77f6e9a52ca 2021-09-11 21:13:55.592864 [DEBUG] switch_rtp.c:7793 Correct audio ip/port confirmed.
9c2dccee-3d82-2451-c81d-ca6bd51c6788 2021-09-11 21:14:30.792863 [DEBUG] switch_rtp.c:1923 rtcp_stats_init: audio ssrc[3520184631] base_seq[9119]
9c2dccee-3d82-2451-c81d-ca6bd51c6788 2021-09-11 21:14:30.792863 [DEBUG] switch_rtp.c:1923 rtcp_stats_init: video ssrc[2887449749] base_seq[13858]
2021-09-11 21:14:30.792863 [DEBUG] mod_verto.c:2045
77.23.137.51:37076 Ending client thread.
---------------------------------------------------------
My understanding of BAD READ is that it usually indicates an incorrectly-created or expired cert, but this BAD READ is happening after the DTLS handshake was completed.
Do you have any recommendation for how we can avoid such BAD READ's?