[TW#27831] esp_mqtt reconnect error: TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected (IDFGH-2975)

[rustyx]

Environment

• Development Kit: ESP32-Wrover-Kit
• Kit version: v4
• Module or chip used: ESP32-WROOM-32
• IDF version: master v3.3-dev-236-gfa59b1b1c
• Build System: Make
• Compiler version: (crosstool-ng-1.22.0-80-g6c4433a5) 5.2.0
• Operating System: Windows 10
• Power Supply: USB

Problem Description

I'm trying out the new esp_mqtt library with an mqtts:// (TSL) connection.

My MQTT server is mosquitto v1.5.3.

The issue is whenever the mqtt server is restarted, the esp_mqtt task hangs with the following messages in the log:

E (56962) TRANS_SSL: esp_tls_conn_read error, errno=Bad file number
E (56967) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (56967) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (56977) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (56982) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (56992) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (56997) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57007) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57012) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57022) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57027) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57037) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57042) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57052) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57062) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57067) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57077) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57082) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57092) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57097) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57107) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57112) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57122) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57127) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57137) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57142) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57152) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57157) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57167) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57172) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57182) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57192) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57197) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57207) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57212) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57222) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57227) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57237) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57242) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57252) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57257) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57267) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57272) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57282) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57287) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57297) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57302) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57312) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57317) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57327) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57337) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57342) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57352) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57357) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57367) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57372) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57382) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57387) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57397) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57402) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57412) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57417) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57427) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57432) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57442) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57447) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57457) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57467) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57472) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57482) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57487) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57497) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57502) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57512) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57517) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57527) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57532) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57542) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57547) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57557) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
E (57562) TRANS_SSL: esp_tls_conn_read error, errno=Socket is not connected
. . .

Other error scenarios like Wi-Fi reconnection, are handled correctly; esp_mqtt is then able to reconnect properly. It is only when the MQTT server connection is closed is it not able to reconnect.

[david-cermak]

Hi @rustyx

It is a bug in ssl transport component. There is already a fix ready in Espressif internal repo, but that still needs to be verified, merged and published to GitHub.

I am attaching a oneliner patch that should address the issue (before it is available in GitHub)
0001-tcp_transport-fixed-ssl-reconnection-issue.patch.txt

Thanks, David

[AadiMehta]

@david-cermak
Thanks david for the patch.

I confirm it works for me.

[david-cermak]

Hi @syed00727

One reason for this kind of error could be a memory corruption.
Also, there's been a recent fix of a race condition on closing mqtt connection here espressif/esp-mqtt@dc1a635. Could you please cherry-pick this commit and check if it resolves the issue?