Comment 2 for bug 2015408

Reviewed: https://review.opendev.org/c/starlingx/snmp-armada-app/+/879689
Committed: https://opendev.org/starlingx/snmp-armada-app/commit/8a6171fce0abdc84fc1488bb78b40a5046fdc4ef
Submitter: "Zuul (22348)"
Branch: master

commit 8a6171fce0abdc84fc1488bb78b40a5046fdc4ef
Author: Agustin Carranza <email address hidden>
Date: Wed Apr 5 17:23:15 2023 -0300

    Refresh DB session after swact in fm-subagent

    After running a swact command, the snmp pod crashed and generated a
    coredump. That behavior was due to an invalid reference for the DB
    handler.
    This change adds a new function to renew the session after the
    first failure (this happens after swact) in the connection.
    During the retries, the session in renewed and the reference is
    restored. So the coredump is now avoided.

    Test plan
    PASS: * Deploy a multinode configuration (e.g. AIO-DX).
          * Install the SNMP application.
          * Perform a snmpget/snmpwalk/snmpbulk operation related to FM,
            to the floating IP, e.g. WRS-ALARM-MIB::wrsAlarmActiveAlarmId.1
          * Perform a swact operation.
          * Repeat the snmp operation.
          * Both operations succeed and no coredumps are found in the
            filesystem of all controllers.

    Closes-bug: 2015408

    Signed-off-by: Agustin Carranza <email address hidden>
    Change-Id: Idf5e8717c4f9605a3a7c1eaee1eb87e38bc305e4