Flow Classifier Validation Error

Bug #1638421 reported by Jong-Geun Park
26
This bug affects 5 people
Affects Status Importance Assigned to Milestone
networking-sfc
Confirmed
Undecided
Priya

Bug Description

Hi,
I have installed networking-sfc mitaka, but when I create the second port-chain, I faced a BadRequest Error as follow example;

BadRequest: Flow Classifier 81471676-b00f-4989-afaa-a33495c4a5a2 conflicts with Flow Classifier 1cf91d4d-6ca8-4ba3-ada6-4c00b72fd5ce in port chain 9c763348-d7f5-425f-859a-aad1d47ed0de.
Neutron server returns request_ids: ['req-47d4abd2-11f9-4ead-8b55-ddf865eb08ba']

From the source code(networking_sfc/db/sfc_db.py),
whenever a new port-chain is created, port-pair-groups and flow-classifiers should be validated whether they conflict with ones used in existing port-chains.

But, during flow classifier validation(function _validate_flow_classifiers),
only basic optional arguments(ether type, protocol, ip prefix and port range) are checked, so it causes the "flow classifier conflict error" with an existing one.

To resolve this bug, in _validate_flow_classifiers,
flowclassifier_basic_conflict need to be replaced with flow classifier_conflict function. (line number 257)

Revision history for this message
Jong-Geun Park (cyberlation) wrote :
Changed in networking-sfc:
status: New → Confirmed
Priya (priyaa)
Changed in networking-sfc:
assignee: nobody → Priya (priyaa)
Revision history for this message
Priya (priyaa) wrote :

 Yes, it should check for other parameters also while validating the flow classifiers.

Revision history for this message
Igor D.C. (igordcard) wrote :

Looks like my ML message relates to this existing bug.
See http://lists.openstack.org/pipermail/openstack-dev/2017-January/109873.html.

More importantly, see http://eavesdrop.openstack.org/meetings/service_chaining/2017/service_chaining.2017-01-12-17.14.log.html for the IRC meeting where we discussed why this is actually not a bug, but expected behavior.

Revision history for this message
Priya (priyaa) wrote :

Yeah, understood this is an expected behavior. Thanks for the info.

Revision history for this message
Bernard Cafarelli (bcafarel) wrote :

That makes me think, we are at least 3 persons who got surprised by this behaviour, it could be worth adding a note somewhere in the documentation about that. I'll bring the question up at next meeting

Revision history for this message
Igor D.C. (igordcard) wrote :

+1

I'm also going to "work around it" for chains that have port pairs with correlation!=None, since the basic conflict check isn't required (and is a blocker) for that use case.

Revision history for this message
murtazasolangi (amurtaza14) wrote :

Also effected with same issue,
"Unable to create flow-classifier for service function chaining: The resource could not be found"

details as per this link: https://ask.openstack.org/en/question/101532/unable-to-create-flow-classifier-for-service-function-chaining-the-resource-could-not-be-found/#

Help will be highly appreciable.

Regards,
Ali

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Bug attachments

Remote bug watches

Bug watches keep track of this bug in other bug trackers.