BGP was down after LM upgrade from 3.1.2.1 to 3.2.0.1
search cancel

BGP was down after LM upgrade from 3.1.2.1 to 3.2.0.1

book

Article ID: 317810

calendar_today

Updated On:

Products

VMware NSX

Issue/Introduction

BGP was down on an NSX local manager , however UI shows the necessary config at edge.
At Edge CLI , get bgp neighbors will not show any output.

Log location :-- manager/controller/data

LM controller data dump for one of the BGP neighbor:—
-----------------------------------
ID: ########-####-####-####-########0401
Type: vmware.nsx.nestdb.BgpNeighborMsg
Receivers: ########-####-####-####-########9257
Datum: id {
  ########-####-####-####-########0401
}
sr_uuid {
  ########-####-####-####-########cad1
}
ip_address {
  ipv4: 10.4.22.17
}
remote_as: 65501
name: "########-####-####-####-########ab8b"
src_ip_address {
  ipv4: 10.4.22.18
}
hold_down_timer: 12
keep_alive_timer: 4
address_family {
  type: IPV4_UNICAST
  enabled: true
  filter_in_prefix_list_uuid {
    ########-####-####-####-########3289
  }
  filter_out_prefix_list_uuid {
    ########-####-####-####-########776b
  }
}
enable_bfd: true
enable: true
max_hop_limit: 1
type: EBGP
neighbor_uuid {
  ########-####-####-####-########76b4
}

——————————————————————————

We don’t see any object for prefix-list :— vmware.nsx.nestdb.PrefixListMsg with uuid ########-####-####-####-########3289 & ########-####-####-####-########776b at data dump.


Edge Logs show the following error:--

2022-02-11T11:20:02.346Z CorpNSXEdge02.jacksonemc.com start-stop-daemon 19673 - - 2022-02-11T11:20:02Z rcpm 19677 rcpm-db [ERROR] prefix_list_in_db Ref Failure: Prefix List (########-####-####-####-########3289) not found in rcpm_db errorCode="EDG0200026"
2022-02-11T11:20:02.587Z CorpNSXEdge02.jacksonemc.com start-stop-daemon 19673 - - 2022-02-11T11:20:02Z rcpm 19677 rcpm-db [ERROR] prefix_list_in_db Ref Failure: Prefix List (########-####-####-####-########776b) not found in rcpm_db errorCode="EDG0200026"


Environment

VMware NSX-T Data Center 3.x

Cause

During local manager onboarding to global manager, we change all intent paths from /infra to /global-infra

In this case , T0 logical router & Prefix-lists object on proton should also updated with tag policypath as /global-infra, but somehow that didn't happen and it has only /infra.

Resolution

This issue is resolved in VMware NSX-T Data Center 3.2.1

Workaround:
Remove the prefix-list applied to the BGP neighbors .

Additional Information

Impact/Risks:
All the traffic from/to local manager towards External world is affected.