[R2.20]DM: Make DM code robust to handle failure scenarios
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R2.20 |
Fix Committed
|
High
|
Suresh Balineni | |||
Trunk |
Fix Committed
|
High
|
Suresh Balineni |
Bug Description
This is a catch-all bug to document a bunch of exception condition handling. DM should be able to handle these failures gracefully.
1. If a commit fails for whatever reason, DM gets stuck and will not attempt to push configs again until its restarted.
2. There could be commit failures if the irb/si-* interfaces created by DM already exist on the MX. DM needs to be able to handle this (1469991).
3. DM tries to do bulk push of large config blobs. This could slow down the netconf process on the MX and cause some configs to be missed. Configs need to be pushed reliably in smaller chunks.
4. There is a timing issue when deleting configs via VNC API. Even after deleting the physical router, the __contrail__ groups config is present on the MX.
5. DM is unaware of the status of the ssh netconf connection with the MX and fails to retry periodically (and sync the config) in the event that the connection goes down (1469366)
description: | updated |
tags: | added: quench |
information type: | Proprietary → Public |
Should fix issue 1, 4 and 5 first so that we can get DM working reliably
for small to medium sized configs.