try to fix broken bgp sessions

[majst01]

fix #7

@mreiger I need to get the exact output of show bgp ipv4 summary json from one machine which is in this situation.

TODO:

• change container to include ethtool again
• access to /var/run/frr/bgpd.vty must be possible

According to frr source, peerstate can be one of:

const struct message bgp_status_msg[] = {{Idle, "Idle"},
					 {Connect, "Connect"},
					 {Active, "Active"},
					 {OpenSent, "OpenSent"},
					 {OpenConfirm, "OpenConfirm"},
					 {Established, "Established"},
					 {Clearing, "Clearing"},
					 {Deleted, "Deleted"},
					 {0}};

[mreiger]

here's the requested output:

shoot--plx7fj--test-vpn-group-0-58bb9-kfgh6# show bgp ipv4 summary json
{
"ipv4Unicast":{
  "routerId":"10.128.144.3",
  "as":4210000113,
  "vrfId":0,
  "vrfName":"default",
  "tableVersion":1451,
  "ribCount":92,
  "ribMemory":16928,
  "peerCount":3,
  "peerMemory":2222424,
  "peerGroupCount":2,
  "peerGroupMemory":128,
  "peers":{
    "10.244.0.15":{
      "dynamicPeer":true,
      "remoteAs":4210000113,
      "localAs":4210000113,
      "version":4,
      "msgRcvd":996855,
      "msgSent":1329966,
      "tableVersion":0,
      "outq":0,
      "inq":0,
      "peerUptime":"04w2d18h",
      "peerUptimeMsec":2658221000,
      "peerUptimeEstablishedEpoch":1655500129,
      "pfxRcd":1,
      "pfxSnt":47,
      "state":"Established",
      "peerState":"OK",
      "connectionsEstablished":1,
      "connectionsDropped":0,
      "idType":"ipv4"
    },
    "lan0":{
      "hostname":"fra-equ01-r01leaf01",
      "remoteAs":4200017001,
      "localAs":4210000113,
      "version":4,
      "msgRcvd":1656139,
      "msgSent":1655620,
      "tableVersion":0,
      "outq":0,
      "inq":0,
      "peerUptime":"6d00h06m",
      "peerUptimeMsec":518789000,
      "peerUptimeEstablishedEpoch":1657639561,
      "pfxRcd":30,
      "pfxSnt":18,
      "state":"Established",
      "peerState":"OK",
      "connectionsEstablished":2,
      "connectionsDropped":1,
      "idType":"interface"
    },
    "lan1":{
      "hostname":"fra-equ01-r01leaf02",
      "remoteAs":4200017002,
      "localAs":4210000113,
      "version":4,
      "msgRcvd":1655341,
      "msgSent":1654501,
      "tableVersion":0,
      "outq":0,
      "inq":0,
      "peerUptime":"00:25:59",
      "peerUptimeMsec":1559000,
      "peerUptimeEstablishedEpoch":1658156791,
      "pfxRcd":0,
      "pfxSnt":0,
      "state":"Idle",
      "peerState":"OK",
      "connectionsEstablished":5,
      "connectionsDropped":5,
      "idType":"interface"
    }
  },
  "failedPeers":1,
  "displayedPeers":3,
  "totalPeers":3,
  "dynamicPeers":1,
  "bestPath":{
    "multiPathRelax":"true"
  }
}
}

[Gerrit91]

Maybe it's easier to merge when we first just crash the node-init container, which will then be reflected in the shoot's system health. The log message could contain a hint on how to resolve the problem?