|
Attachment overcloud-controller-0-karaf-try4.zip has been added with description: logs fron controller-0
|
|
Attachment overcloud-controller-1-karaf-try4.zip has been added with description: logs fron controller-1
|
|
Attachment overcloud-controller-2-karaf-try4.zip has been added with description: logs fron controller-2
|
|
Attachment overcloud-controller-0-karaf-more-logs.log has been added with description: Logs from controller-0 after adding more logging into code
|
|
Attachment overcloud-controller-1-karaf-more-logs.zip has been added with description: Logs from controller-1 after adding more logging into code
|
|
Attachment overcloud-controller-2-karaf-more-logs.zip has been added with description: Logs from controller-2 after adding more logging into code
|
|
It seems, that the actual problem we are hitting here is that custom schema-cache-directory is not working correctly in clustered setting.
Mounting a netconf device in clustered setting goes as follows. Leader for netconf mounpoint is elected (this is done with help of clustering singleton service), leader mounts a device and notifies followers about mounting result. Part of the leader's mount process is device schema resolution.
Leader downloads all the schema sources from device, registers them with its schema registry and schema repository and finally builds schema context. After that, leader sends followers list of supported source ids and also creates actor that serves the sources upon followers' requests. It seems, that this actor is injected with wrong schema repository so it cannot serve followers requests and we see this failures in the log.
The bug should be present in a setting where we specify a custom schema-cache-directory for a netconf mount. Default setting shouldn't cause any problems.
|
|
When using this configuration:
{
"topology": [
{
"node": [
{
"netconf-node-topology:actor-response-wait-time": 20,
"netconf-node-topology:between-attempts-timeout-millis": 10000,
"netconf-node-topology:connection-timeout-millis": 10000,
"netconf-node-topology:default-request-timeout-millis": 10000,
"netconf-node-topology:host": "192.168.10.25",
"netconf-node-topology:keepalive-delay": 0,
"netconf-node-topology:max-connection-attempts": 10,
"netconf-node-topology:password": "admin",
"netconf-node-topology:port": 2831,
"netconf-node-topology:tcp-only": false,
"netconf-node-topology:username": "admin",
"node-id": "overcloud-controller-0.opnfv.org"
}
],
"topology-id": "topology-netconf"
}
]
}
the errors disappeared. The errors were present only when I used a custom schema-cache-directory, e.g. "netconf-node-topology:schema-cache-directory": "hcmount"
|
|
https://git.opendaylight.org/gerrit/#/c/54017/
|
|
A suite attempting to detect this Bug has been merged [0].
But I was unable to find a build where the suite fails.
The suite uses netconf-testtool and manipulates device data modeled by car.yang which is both exported by testtool and present in ODL-local /tmp/schema directory (on every cluster member).
Please, let me know if you find a build which fails on the suite, or if you spot a difference between the scenario the suite is testing and original Bug description.
[0] https://git.opendaylight.org/gerrit/54110
|
Generated at Wed Feb 07 20:14:53 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.
overcloud-controller-0-karaf-more-logs.log
overcloud-controller-0-karaf-try4.zip