Manuel,

About the first problem, I'll have to check, but I think we need to consider 
the metadata in the match, and the metadata in the action. Notice the match 
"metadata=0x10000000000/0xffffff0000000000" has the value 1 as expected, since 
this particular flow is for SFC and goes to table 83 which is the SFC SFF. The 
metadata set in the action is most likely for the next service after SFC. So, 
in this flow the packet will be sent to SFC, who will process it and when/if it 
gets resubmit back to table 17, the next service to be invoked will be L3VPN as 
indicated by "write_metadata:0x8000010000000000/0xfffffffffffffffe" in the SFC 
flow in table 17.


About the second problem, it sounds like the packets arent classified 
correctly. That is, if this flow gets hit:

 table=83, n_packets=338, n_bytes=37768, priority=5 actions=drop

And these 2 are not hit:

 table=83, n_packets=0, n_bytes=0, priority=250,nsp=11 actions=goto_table:86
 table=83, n_packets=0, n_bytes=0, priority=250,nsp=8388619 
actions=goto_table:86

Then the packets dont have NSH and the packets get encapsulated with NSH by the 
classifier.

Regards,

Brady


-----Original Message-----
From: Manuel Buil <[email protected]<mailto:manuel%20buil%20%[email protected]%3e>>
To: [email protected] 
<[email protected]<mailto:%[email protected]%22%20%[email protected]%3e>>
Cc: Kency Kurian 
<[email protected]<mailto:kency%20kurian%20%[email protected]%3e>>,
 Faseela K 
<[email protected]<mailto:faseela%20k%20%[email protected]%3e>>
Subject: [sfc-dev] Two potential bugs in Carbon SFC?
Date: Thu, 20 Apr 2017 15:13:46 +0200

Hey guys,

I am using a build from Monday and I found two things which might be wrong. 
Before adding them to bugzilla, I wanted to check with you.

### Number 1 ### @Faseela, Kency here you might help :)

When creating the RSP this flow is created in table=17 of the SFF:

table=17, n_packets=6, n_bytes=252, 
priority=10,metadata=0x10000000000/0xffffff0000000000 
actions=write_metadata:0x8000010000000000/0xfffffffffffffffe,goto_table:83

As far as I understood, the first 4 bits of the metadata should point to the id 
of the service and here it points to id = 8 which according to this file in 
genius:

https://git.opendaylight.org/gerrit/gitweb?p=genius.git;a=blob;f=mdsalutil/mdsalutil-api/src/main/java/org/opendaylight/genius/mdsalutil/NwConstants.java;h=9b2a2a94283684c8e56a9a53df32b48d878dd83a;hb=refs/heads/master

id = 8 ==> L3VPN_SERVICE_INDEX

It should be SFC_SERVICE_INDEX = 1; right?


### Number 2 ###

After creating the RSP all my packets are dropped at table=83:

 table=83, n_packets=0, n_bytes=0, priority=250,nsp=11 actions=goto_table:86
 table=83, n_packets=0, n_bytes=0, priority=250,nsp=8388619 
actions=goto_table:86
 table=83, n_packets=338, n_bytes=37768, priority=5 actions=drop

Instead of getting dropped, they should go back to table=17 so that they are 
moved to the next registered service or? For example, pinging with the client 
to the SFF does not work because of this.

Thanks,
Manuel

_______________________________________________
sfc-dev mailing list
[email protected]<mailto:[email protected]>
https://lists.opendaylight.org/mailman/listinfo/sfc-dev

_______________________________________________
sfc-dev mailing list
[email protected]
https://lists.opendaylight.org/mailman/listinfo/sfc-dev

Reply via email to