[fdbshow]: Handle FDB cleanup gracefully

[nazariig]

Signed-off-by: Nazarii Hnydyn nazariig@nvidia.com

The race condition is caused by a blocking Redis call which gets the contents of the FDB entry from ASIC DB.
Since it has been implemented as a simple loop, there is no guarantee that entry will be present in DB when the contents are being read.

root@sonic:/home/admin# ./reproduce.sh
...
Tue Nov  9 15:58:49 UTC 2021: Query...
Tue Nov  9 15:58:52 UTC 2021: Query...
...
=> Read ASIC DB: start
=> Read ASIC DB: done
=> Read ASIC DB: start
Key 'ASIC_STATE:SAI_OBJECT_TYPE_FDB_ENTRY:{"bvid":"oid:0x260000000009cc","mac":"02:11:22:33:20:00","switch_id":"oid:0x21000000000000"}' unavailable in database '1'

The fdbshow script:

root@sonic:/home/admin# vim /usr/bin/fdbshow

...

class FdbShow(object):

    HEADER = ['No.', 'Vlan', 'MacAddress', 'Port', 'Type']
    FDB_COUNT = 0

    def __init__(self):
        super(FdbShow,self).__init__()
        self.db = SonicV2Connector(host="127.0.0.1")
        self.if_name_map, \
        self.if_oid_map = port_util.get_interface_oid_map(self.db)
        self.if_br_oid_map = port_util.get_bridge_port_map(self.db)
        self.fetch_fdb_data()
        return

    def fetch_fdb_data(self):
        """
            Fetch FDB entries from ASIC DB.
            FDB entries are sorted on "VlanID" and stored as a list of tuples
        """
        self.db.connect(self.db.ASIC_DB)
        self.bridge_mac_list = []

        fdb_str = self.db.keys('ASIC_DB', "ASIC_STATE:SAI_OBJECT_TYPE_FDB_ENTRY:*")
        if not fdb_str:
            return

        if self.if_br_oid_map is None:
            return

        oid_pfx = len("oid:0x")
        for s in fdb_str:
            fdb_entry = s.decode()
            fdb = json.loads(fdb_entry .split(":", 2)[-1])
            if not fdb:
                continue

            print("=> Read ASIC DB: start")
            ent = self.db.get_all('ASIC_DB', s, blocking=True)
            print("=> Read ASIC DB: done")
            br_port_id = ent[b"SAI_FDB_ENTRY_ATTR_BRIDGE_PORT_ID"][oid_pfx:]
            ent_type = ent[b"SAI_FDB_ENTRY_ATTR_TYPE"]
            fdb_type = ['Dynamic','Static'][ent_type == "SAI_FDB_ENTRY_TYPE_STATIC"]

...

The reproduce.sh script:

root@sonic:/home/admin# cat reproduce.sh
#!/bin/bash

out=""
rc="0"

while [[ "${rc}" -eq "0" ]]; do
    echo "$(date): Query..."
    out="$(show mac)"
    rc="$?"
done

echo "${out}"

Closes #1866

What I did

• Fixed: [fdb] 'show mac' command failed with t0-56-po2vlan topology #1866

How I did it

• Removed blocking calls from fdbshow

How to verify it

1. Run FDB test

Which release branch to backport (provide reason below if selected)

• 201911
• 202012
• 202106

Previous command output (if the output of a command-line utility has changed)

root@sonic:/home/admin# show mac
Key 'ASIC_STATE:SAI_OBJECT_TYPE_FDB_ENTRY:{"bvid":"oid:0x260000000009cc","mac":"02:11:22:33:20:00","switch_id":"oid:0x21000000000000"}' unavailable in database '1'

New command output (if the output of a command-line utility has changed)

root@sonic:/home/admin# show mac
No.    Vlan    MacAddress    Port    Type
-----  ------  ------------  ------  ------
Total number of entries 0

[nazariig]

#1918

[liat-grozovik]

@prsunny could you please help to review it?

[nazariig]

/azpw run Azure.sonic-utilities (Test vstest)

[mssonicbld]

/AzurePipelines run Azure.sonic-utilities (Test vstest)

[azure-pipelines]

No pipelines are associated with this pull request.

[nazariig]

/azpw run Azure.sonic-utilities vstest

[mssonicbld]

/AzurePipelines run Azure.sonic-utilities vstest

[azure-pipelines]

No pipelines are associated with this pull request.

[nazariig]

/azpw run Azure vstest

[mssonicbld]

/AzurePipelines run Azure vstest

[azure-pipelines]

No pipelines are associated with this pull request.

[nazariig]

/azpw run

[mssonicbld]

/AzurePipelines run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[nazariig]

@liat-grozovik can you please help to merge?