[NETCONF-128] Cannot detect netconf up and running when odl-netconf-connector-* is not used Created: 19/Jan/16  Updated: 05/Dec/22  Resolved: 22/Jan/16

Status: Resolved
Project: netconf
Component/s: netconf
Affects Version/s: None
Fix Version/s: None

Type: Bug
Reporter: Jozef Behran Assignee: Tomas Cere
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All

Issue Links:
Relates
relates to NETCONF-919 Do not touch configuration datastore ... Resolved
External issue ID: 5014

 Description   

When odl-netconf-connector-* feature is not installed (example: Netconf clustered CRUD tests [1], [2], [3], [4]), then there is no way to remotely detect whether Netconf is ready or not (and when that happens). Locally it is possible to grep logs for "<some netconf related component> is up/initialized successfully/whatever" message but when no ssh access is provided, this can't help much. This currently breaks "netconf readiness" suite as it was intended to detect exactly when Netconf is ready.

Investigation revealed that the problem is caused by bizzare behavior of Netconf. All Netconf related Restconf URLs will return "404 Not Found" until a Netconf device is configured, then they mysteriously start to report "200 Ok". Tested on Netconf topology. The correct behavior (at least for topology) is to start returning "200 Ok" as soon as devices can be configured.

[1] https://jenkins.opendaylight.org/releng/view/netconf/job/netconf-csit-3node-clustering-only-beryllium/
[2] https://jenkins.opendaylight.org/releng/view/netconf/job/netconf-csit-3node-clustering-all-beryllium/
[3] https://jenkins.opendaylight.org/releng/view/netconf/job/netconf-csit-3node-clustering-only-boron/
[4] https://jenkins.opendaylight.org/releng/view/netconf/job/netconf-csit-3node-clustering-all-boron/

 Comments   
Comment by Jozef Behran [ 19/Jan/16 ]

This change is reportedly fixing this problem: https://git.opendaylight.org/gerrit/#/c/32936/

Comment by Jozef Behran [ 19/Jan/16 ]

It appears the patch above was merged but is still not in the artifacts in Nexus. So a multipatch-sandbox-build job is due to be done to make sure this bug is not there. Changing to "waiting for review".

Comment by Jozef Behran [ 20/Jan/16 ]

Just completed testing on a Beryllium snapshot (had to build it locally as the build server(s) are broken; build was made from stable/beryllium sources as they were yesterday evening). This bug is still present.

Comment by Jozef Behran [ 20/Jan/16 ]

(changed to "CONFIRMED" as the bug is confirmed to survive the patches)

Comment by Jozef Behran [ 21/Jan/16 ]

Another change that contributes to fixing this bug: https://git.opendaylight.org/gerrit/32939/

Comment by Jozef Behran [ 21/Jan/16 ]

Results of discussion with developers:

1. The fixes add the topology into the operational datastore.
However the test suite is searching it in config datastore.
We agreed that it would be good to add it into the config datastore
as well as some components might search it there and get confused
if not found.
2. The fixes did not make it to beryllium before I built my image so I
did not have them (yet).

Going to retest with the latest image of netconf. Right now leaving at "confirmed" as the bug is still there (the netconf may be invisible to components which search for topology in config datastore only).

Comment by Tomas Cere [ 21/Jan/16 ]

https://git.opendaylight.org/gerrit/#/c/33196/

Generated at Wed Feb 07 20:14:14 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.