The SSL_shutdown() errors reported are likely to be expected on connection timeout where, while shutting down the SSL layer on the connection, some data is then received from the client. This means that the application layer of the connection isn't properly terminated (as due to timeout). Since the stream module doesn't usually involve on the application layer, these errors can be interpreted as innocent.
A common suggestion would be to adjust timeouts (see the proxy_timeout directive as the most suitable in the stream module).
You can apply this patch to lower the logging level of such errors:
# HG changeset patch
# User Sergey Kandaurov <firstname.lastname@example.org>
# Date 1644330927 -10800
# Tue Feb 08 17:35:27 2022 +0300
# Node ID a736a7a613ea6e182ff86fbadcb98bb0f8891c0b
# Parent 1add55d236522616ce34ffaa4dc697a76d3d41a4
SSL: logging level of "application data after close notify".
Such fatal errors are reported by OpenSSL 1.1.1, and similarly by BoringSSL,
if application data is encountered during SSL shutdown, which started to be
observed on the second SSL_shutdown() call after SSL shutdown fixes made in
09fb2135a589 (1.19.2). The error means that the client continues to send
application data after receiving the "close_notify" alert (ticket #2318).
Previously it was reported as SSL_shutdown() error of SSL_ERROR_SYSCALL.
diff --git a/src/event/ngx_event_openssl.c b/src/event/ngx_event_openssl.c
@@ -3385,6 +3385,12 @@ ngx_ssl_connection_error(ngx_connection_
|| n == SSL_R_WRONG_VERSION_NUMBER /* 267 */
|| n == SSL_R_DECRYPTION_FAILED_OR_BAD_RECORD_MAC /* 281 */
+ || n == SSL_R_APPLICATION_DATA_AFTER_CLOSE_NOTIFY /* 291 */
+ || n == SSL_R_APPLICATION_DATA_ON_SHUTDOWN /* 291 */
|| n == SSL_R_RENEGOTIATE_EXT_TOO_LONG /* 335 */
|| n == SSL_R_RENEGOTIATION_ENCODING_ERR /* 336 */