[CONTROLLER-1259] Error messages and logs missing for this or other RPC failures. Created: 17/Apr/15  Updated: 28/May/15  Resolved: 28/May/15

Status: Resolved
Project: controller
Component/s: restconf
Affects Version/s: Post-Helium
Fix Version/s: None

Type: Bug
Reporter: Phillip Shea Assignee: Vaclav Demcak
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All

External issue ID: 3017

 Description   

This is related to CONTROLLER-1258. In addition to the problem encountered in the bug, it lacked a useful error message and log entry. This bug should be fixed first, since the other bug is needed to reproduce it.

I'm encountered this running /suites/clustering/datastore/020_crud_on_leader.txt.

The Keywords that are failing are:

Purchase cars on Leader
Purchase cars on Follower1
Purchase cars on Follower2

What's happening is that the RPCs to "buy a car" only work when executed against one of the three controllers. Strangely it's not even the car-people shard leader.

Tom Pantelis is running these test in Brocade's lab on Helium and they are passing. This bug is unique to Lithium.

Other than the fact that these keywords are failing, this bug addresses the issue that the error messages and log entries are missing when the "buy a car" RPCs are used.

Procedure:

Pull the latest stable Lithium Release. I pulled distribution-karaf-0.3.0-20150414.220637-1140.zip

Use the python cluster deploy script.

Let the test run. The shards configured themselves in this way:

car shard

{'member1': 'Follower', 'member2': 'Leader', 'member3': 'Follower'}

people shard

{'member1': 'Follower', 'member2': 'Follower', 'member3': 'Leader'}

car-people shard

{'member1': 'Leader', 'member2': 'Follower', 'member3': 'Follower'}

After the test completes, use ReST RPCs to purchase a car on member1, member2, and member2.

URL:
http://member1:8181/restconf/operations/car-purchase:buy-car

JSON:
{ "input" :

{ "car-purchase:person" : "/people:people/people:person[people:id='user9']", "car-purchase:person-id" : "user9", "car-purchase:car-id" : "9" }

}

For the first two members, there will be a 500 error in Postman, and nothing in the karaf log. Member3 gets a nice “200 OK” in Postman and I see an acknowledgement of the car purchase in the karaf log.

The 500 error should contain some text (or even a stack trace - if that's not a security issue) describing the reason for the error. In this case the 500 error has no text.

To observe the log files connect to the terminal on all three controllers.

ssh karaf@member1 -p 8101
ssh karaf@member2 -p 8101
ssh karaf@member3 -p 8101

If you try the "buy a car" RPCs, you will see nothing from the failures on member1 and member2, but you will see a acknowledgement message for the successful RPC on member3.

The failures should produce both errors in the RPC ReST call and in the karaf log.

 Comments   
Comment by Tom Pantelis [ 17/Apr/15 ]

This is about reporting detailed error info on rpc failure. The error message just says something "The operation encountered an unexpected error while executing." with no other info:

{
"errors":{
"error":[

{ "error-type":"application", "error-tag":"operation-failed", "error-message":"The operation encountered an unexpected error while executing." }]}

The error-info tag should be included with the detailed error cause, preferably including the stack trace. As I recall, it used to include this.

As long as we return the detailed error info, I don't think we need to log it as warn or error in the karaf log.

Comment by Vaclav Demcak [ 15/May/15 ]

Restconf doesn't contain logs for more functionality (not only RPC)
so next patch adds better logging for Restconf (BrokerFacade and RestconfImpl)

https://git.opendaylight.org/gerrit/#/c/19712/

Generated at Wed Feb 07 19:55:05 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.