Add LocalSecGrpClassifierCounter #581
Conversation
mchalla
commented
Aug 1, 2024
mchalla
commented
- added as an extension of SecGrpClassifierCounter
- added make checks as a copy of SecGrpStatsManager_test.cpp, replacing all the classifiers with Local, and convert the called functions to use template arguments.
- The make checks cover the prometheus tests
- this would also result in pushing the LocalSecGrpClassifierCounter to the leaf, should we decide to push the local secgrps to the leaf, we can strip the Local before sending it.
|
I missed changing testSecGrpDelete to use local classifier. I am seeing an issue where deleting the classifier is not deleting the counter for local case and will update with a fix. |
- added as an extension of SecGrpClassifierCounter - added make checks as a copy of SecGrpStatsManager_test.cpp, replacing all the classifiers with Local, and convert the called functions to use template arguments. - PolicyStatsManager also listens to LocalL24Classifiers - The make checks cover the prometheus tests - this would also result in pushing the LocalSecGrpClassifierCounter to the leaf, should we decide to push the local secgrps to the leaf, we can strip the Local before sending it. The following is the mapping L24Classifier <Contract> L24ClassifierCounter L24Classifier <SecGroup> SecGrpClassifierCounter LocalL24Classifier <LocalSecGroup> LocalSecGrpClassifierCounter Signed-off-by: Madhu Challa <challa@gmail.com>
mchalla
force-pushed
the
local-hpp-counters
branch
from
37a90a3
to
4b0b5d1
Compare
The issue was the policy stats manager not listening to locall24classifier events. After fixing it the test passes. |
| @@ -46,7 +47,8 @@ using boost::system::error_code; | |||
| SecGrpStatsManager::SecGrpStatsManager(Agent* agent_, IdGenerator& idGen_, | |||
| SwitchManager& switchManager_, | |||
| long timer_interval_) | |||
| : PolicyStatsManager(agent_,idGen_,switchManager_,timer_interval_) { | |||
| : PolicyStatsManager(agent_,idGen_,switchManager_,timer_interval_), | |||
| agent(agent_) { | |||
There was a problem hiding this comment.
Why do you need to store Agent* in the SecGrpStatsManager subclass as well. It's protected in the superclass. Can't you just access it from there as needed?
There was a problem hiding this comment.
oh i did not realize that. i will change it via a separate commit since i already merged this change.
