Hi !
As the vQFX is not officially supported here is no other way like j-TAC to inform about bugs
Here is a nice bug I have found during my extensive testing of various DC-configurations :
Description: On vQFX10k 15.1X53-D60 when configuring BOTH family evpn signalling AND family route-target the system only advertises route-target and NO EVPN route
When deactivating family route-target the EVPN routes will get advertised
PROOF:
lab@QFX51> show configuration protocols bgp group EVPN-OVERLAY
type internal;
local-address 10.0.0.51;
family inet-vpn {
unicast;
}
family evpn {
signaling;
}
family route-target;
multipath;
neighbor 10.0.0.5;
result.:
lab@QFX51> show route advertising-protocol bgp 10.0.0.5
bgp.rtarget.0: 6 destinations, 7 routes (1 active, 0 holddown, 6 hidden)
Prefix Nexthop MED Lclpref AS path
65501:65501:99/96
* Self 100 I
{master:0}
---> only route target is announced
lab@QFX51> show route table default-switch.evpn.0 extensive
2:10.0.0.52:1::1051::02:05:86:71:e8:00/304 (1 entry, 1 announced)
TSI:
RTF { 65501:99/64 }
*EVPN Preference: 170
Next hop type: Indirect, Next hop index: 0
Address: 0x9db9df0
Next-hop reference count: 50
Protocol next hop: 10.0.0.52
Indirect next hop: 0x0 - INH Session ID: 0x0
State: <Active Int Ext>
Age: 41:15
Validation State: unverified
Task: default-switch-evpn
Announcement bits (1): 2-BGP Route Target
AS path: I
Communities: encapsulation0:0:0:0:vxlan evpn-default-gateway
Route Label: 1051
ESI: 00:00:00:00:00:00:00:00:00:00
############################## disabling family route target --->
lab@QFX51> show route advertising-protocol bgp 10.0.0.5
default-switch.evpn.0: 78 destinations, 78 routes (78 active, 0 holddown, 0 hidden)
Prefix Nexthop MED Lclpref AS path
2:10.0.0.51:1::1050::00:00:5e:00:01:01/304
* Self 100 I
2:10.0.0.51:1::1050::02:05:86:71:25:00/304
* Self 100 I
2:10.0.0.51:1::1051::00:00:5e:00:01:01/304
* Self 100 I
2:10.0.0.51:1::1051::02:05:86:71:25:00/304
* Self 100 I
2:10.0.0.51:1::1052::00:00:5e:00:01:01/304
* Self 100 I
2:10.0.0.51:1::1052::02:05:86:71:25:00/304
* Self 100 I
2:10.0.0.51:1::1090::00:00:5e:00:01:01/304
* Self 100 I
2:10.0.0.51:1::1090::02:05:86:71:25:00/304
.......
Another "BUG" is seen on the 10.0.0.5 which plays the role of a ROUTE REFLECTOR
(vMX 14.1R1.10)
instructor@MX155-RR> show route table bgp.evpn.0
bgp.evpn.0: 83 destinations, 83 routes (0 active, 0 holddown, 83 hidden)
1:10.0.0.51:0::050000ffdd0000041c00:ffff:65535/304 (1 entry, 0 announced)
BGP Preference: 170/-101
Route Distinguisher: 10.0.0.51:0
Next hop type: Unusable
Address: 0x9293e84
Next-hop reference count: 83
State: <Hidden Int Ext>
Local AS: 65501 Peer AS: 65501
Age: 4:33
Validation State: unverified
Task: BGP_65501.10.0.0.51+53139
AS path: I
Communities: target:65501:99 unknown iana 30c esi-label:000000(label 0)
Accepted
Localpref: 100
Router ID: 10.0.0.51
Indirect next hops: 1
Protocol next hop: 10.0.0.51
Indirect next hop: 0x0 - INH Session ID: 0x0
---> it shows the next-hop as unsuable even though there is a route in inet.0
inet.0: 59 destinations, 59 routes (59 active, 0 holddown, 0 hidden)
+ = Active Route, - = Last Active, * = Both
...
10.0.0.51/32 *[OSPF/10] 1w0d 20:46:17, metric 9999
> to 10.5.51.51 via ge-0/0/1.51
...
even though the EVPN/VXLAN route does not need any LSP it tries to resolve in table inet.3
where this route is missing
but when configuring a static route for inet.3
set rib inet.3 static route 10.0.0.0/24 next-hop 10.0.0.6 resolve
then
instructor@MX155-RR> show route table inet.3
inet.3: 6 destinations, 6 routes (6 active, 0 holddown, 0 hidden)
+ = Active Route, - = Last Active, * = Both
10.0.0.0/24 *[Static/5] 00:18:30, metric2 1
> to 10.5.6.6 via ge-0/0/3.5
and the routes are not hidden anymore
instructor@MX155-RR> òòòshow route table bgp.evpn
bgp.evpn.0: 84 destinations, 84 routes (84 active, 0 holddown, 0 hidden)
+ = Active Route, - = Last Active, * = Both
1:10.0.0.51:0::050000ffdd0000041a00:ffff:65535/304
*[BGP/170] 00:09:39, localpref 100, from 10.0.0.51
AS path: I, validation-state: unverified
> to 10.5.6.6 via ge-0/0/3.5
..
########################################################################
So hopefully Juniper will have a look into that things and sort it out, I DO not know if the behaviour is the same on physical QFX and MX and on MX in newer versions
########################################################################
With best regards
Alexander Marhold
JNCIP-DC