Hey stunnel-users,
We are using stunnel as the encryption transit between our NFS client and server. Recently we detected that the stunnel aborted the workflow silently in half way for unknown reason. We'd like to understand what could be the potential reason that cause the issue.
Stunnel version: 4.56/5.56, below logs are for 4.56, 5.56 is from customer report without detail log
Occurance #1:
2022.03.17 02:19:58 LOG7[1105:139990955783936]: connect_blocking: s_poll_wait IP:PORT: waiting 10 seconds
# -> There should be a `TIMEOUTconnect exceeded` failure if the connection failed, but there is no info about how is the waiting going on, there is no log further
2022.03.17 02:56:09 LOG7[1105:139990955789440]: Dispatching signals from the signal pipe
Occurance #2:
2022.03.23 16:10:49 LOG6[19003:140551554406144]: CERT: Host name "**" matched with "**"
# Aborted again, this is the handshake phase
2022.03.23 16:56:09 LOG7[19003:140551554411648]: Dispatching signals from the signal pipe
From the PCAP of the #2 occurance, there is no client side cert and key send to server side, so we can rule out the network issue.
8510 2022-03-23 16:09:47.660910 A B TLSv1.2 5507 Server Hello, Certificate[Packet size limited during capture]
8511 2022-03-23 16:09:47.660919 B A TCP 68 38390 → 2049 [ACK] Seq=340 Ack=5440 Win=57472 Len=0 TSval=698509941 TSecr=1821371384
... No other client side packet send to server side afterwards, server side are waiting
Does anyone encounter this before? I searched the archive but there is no similar issue. What could cause stunnel swallow all the subsequent workflow?
Thanks.