David Lamparter [Tue, 18 Jun 2019 04:15:21 +0000 (06:15 +0200)]
FRRouting release 7.1
- gRPC northbound plugin
- "table NNN" removed from zebra
- more dataplane MT work
- EVPN in non-default VRFs
- RFC 8212 (default deny policy for eBGP)
- RFC 8106 (IPv6 RA DNS options)
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Sun, 9 Jun 2019 23:35:04 +0000 (01:35 +0200)]
tools: retain sanity when reloading under systemd
Without this, we end up restarting watchfrr with the systemd watchdog
non-functional & tripped a bit later. Also, if watchfrr is in the
"control" cgroup, systemd 232 will kill it. (241 apparently doesn't.
Can't find anything about this in systemd's ChangeLog though.)
Donald Sharp [Wed, 5 Jun 2019 01:38:11 +0000 (21:38 -0400)]
bfdd: Modify bfdd to quietly accept access-lists
The `access-list ...` command was causing bfdd to return
'unknown commands'. Make bfdd at least cognizant of
access-lists enough to not create strange error messages
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Donatas Abraitis [Mon, 20 May 2019 13:43:01 +0000 (16:43 +0300)]
bgpd: Show FQDN in `show [ip] bgp` output
We already show this information in `show [ip] bgp <prefix`, thus why don't
show it in global output. It's very handy when using at scale and to see
the whole picture instead of resolving neighbor manually.
It will show FQDN only if `bgp default show-hostname` is toggled.
Martin Winter [Mon, 18 Feb 2019 15:57:24 +0000 (07:57 -0800)]
snapcraft: Disable libyang CACHE for snap package (Temp workaround)
There seems to be a bug in latest libyang running when running in
a snap container which causes a failure of the cache logic. Disable
CACHE for now as it's not yet needed. Will be re-enabled in a later
release
Signed-off-by: Martin Winter <mwinter@opensourcerouting.org>
David Lamparter [Mon, 20 May 2019 21:41:16 +0000 (23:41 +0200)]
bgpd: use DLIST for adv_fifo
I mistakenly assumed that something called "FIFO" would primarily be
used by removing items from the front. This isn't the case for the
adv_fifo... so use a DLIST there to make sure deletions from the middle
of the queue are fast.
Issue 1:
Getting an empty json without any warning message, while executing
the command "show ip bgp neighbor <x.x.x.x> advertised-routes
json" when the bgp instance is not present or getting created.
Issue 2:
Getting an empty json without any warning message, while executing
the command "show ip bgp vrf/view <name> advertised-routes json"
when the specified view/vrf is not present.
Fix:
Display warning message while executing the above cli commands, when
the bgp instance, specified vrf is not present.
Renato Westphal [Mon, 13 May 2019 18:18:57 +0000 (15:18 -0300)]
lib: fix crash in the CLI grammar sandbox
The CLI grammer sandbox needs to initialize the northbound subsystem
otherwise the running_config global variable won't be set, which
leads to crashes.
Bug: If there are 2 different prefixes under an rd, the
output of "show bgp l2vpn evpn json" would print only one of the
prefixes.
RCA: prefix info was added to the json_object once per rd. Hence,
prefix and rd were added just once, as the loop iterated over all
the prefixes and paths.
This is related to my earlier commit that went in via PR 4283:
https://github.com/FRRouting/frr/pull/4283
Chirag Shah [Thu, 9 May 2019 19:25:44 +0000 (12:25 -0700)]
zebra: unset sticky mac upon local deletion
if the local sticky mac delete request is received,
if there are associated neighbor entries present, mac's
only local flag is removed and marked as auto mac.
this results in next local mac learning automatically assumes
mac is sticky.
There is a case when bridge learning off is configured, user
configures sticky mac via bridge fdb add.
This MAC learns associated neighbor entry.
Later user deletes stick mac via bridge fdb del, this triggers
frr to delete mac but if there are neighbors present, frr marks
MAC as AUTO but does not remove sticky flag.
User enables bridge learning on which triggers
The mac to learn as dynamic entry and in absence of this
fix, the mac is marked as sticky.
Issue1: The following json cli commands are broken (they dont display
anything):
1. show [ip] bgp l2vpn evpn json
2. show [ip] bgp l2vpn evpn all neighbors <IP> routes json
3. show [ip] bgp l2vpn evpn rd <RD> json
4. show [ip] bgp l2vpn evpn rd <RD> neighbors <IP> routes json
RCA1: The existing json_object was partially filled.
Some fields like "prefix", "prefixLen", etc were not
filled in the json_object.
RCA2: Code missing to display the constructed json_object.
Issue2: Column header not shown on bgp evpn commands.
Fix: turned on the flag to display header
Output after fix:
The above commands have been made to follow the existing output style of
"show bgp l2vpn evpn routes json"
leaf-1# sh bgp l2vpn evpn json
{
"bgpTableVersion":1,
"bgpLocalRouterId":"10.100.0.1",
"defaultLocPrf":100,
"localAS":65000,
"10.100.0.1:3":{
"rd":"10.100.0.1:3",
"[3]:[0]:[32]:[10.100.0.1]":{
"prefix":"[3]:[0]:[32]:[10.100.0.1]",
"prefixLen":288,
"paths":[
{
"valid":true,
"bestpath":true,
"pathFrom":"external",
"routeType":3,
"ethTag":0,
"ipLen":32,
"ip":"10.100.0.1",
"weight":32768,
"peerId":"(unspec)",
"aspath":"",
"path":"",
"origin":"IGP",
"nexthops":[
{
"ip":"10.100.0.1",
"afi":"ipv4",
"used":true
}
]
}
]
}
},
..........
..........
"10.100.0.2:2":{
"rd":"10.100.0.2:2",
"[3]:[0]:[32]:[10.100.0.2]":{
"prefix":"[3]:[0]:[32]:[10.100.0.2]",
"prefixLen":288,
"paths":[
{
"valid":true,
"bestpath":true,
"pathFrom":"external",
"routeType":3,
"ethTag":0,
"ipLen":32,
"ip":"10.100.0.2",
"weight":0,
"peerId":"205.0.113.2",
"aspath":"65002",
"path":"65002",
"origin":"IGP",
"nexthops":[
{
"ip":"10.100.0.2",
"afi":"ipv4",
"used":true
}
]
},
{
"valid":true,
"pathFrom":"external",
"routeType":3,
"ethTag":0,
"ipLen":32,
"ip":"10.100.0.2",
"weight":0,
"peerId":"203.0.113.4",
"aspath":"65001 65002",
"path":"65001 65002",
"origin":"IGP",
"nexthops":[
{
"ip":"10.100.0.2",
"afi":"ipv4",
"used":true
}
]
}
]
}
},
....
....
"numPrefix":10,
"totalPrefix":10
}
leaf-1# sh bgp l2vpn evpn
BGP table version is 1, local router ID is 10.100.0.1
Status codes:s suppressed, d damped, h history, * valid, > best, i - int
Origin codes: i - IGP, e - EGP, ? - incomplete
Network Next Hop Metric LocPrf Weight Path
Route Distinguisher: ip 10.100.0.1:3
*> [3]:[0]:[32]:[10.100.0.1]
10.100.0.1 32768 i
Route Distinguisher: ip 10.100.0.1:4
*> [3]:[0]:[32]:[10.100.0.1]
10.100.0.1 32768 i
Route Distinguisher: ip 10.100.0.1:5
*> [3]:[0]:[32]:[10.100.0.1]
10.100.0.1 32768 i
Route Distinguisher: ip 10.100.0.1:6
*> [3]:[0]:[32]:[10.100.0.1]
10.100.0.1 32768 i
Route Distinguisher: ip 10.100.0.2:2
*> [3]:[0]:[32]:[10.100.0.2]
10.100.0.2 0 65002 i
* [3]:[0]:[32]:[10.100.0.2]
10.100.0.2 0 65001 65002
Route Distinguisher: ip 10.100.0.2:3
*> [3]:[0]:[32]:[10.100.0.2]
10.100.0.2 0 65002 i
* [3]:[0]:[32]:[10.100.0.2]
10.100.0.2 0 65001 65002
Route Distinguisher: ip 10.100.0.2:4
Root case :
-----------
When RP acts as a LHR i.e RP has a local receiver and registed for
the same group where LHR connected receiver also registered for the
same multicast group.When RP receives a (s,g) join form LHR , it
increments upstream ref count to two to track the Local membership
as well.But at the time of KAT expiry in RP , upstream reference
is not being removed Which is added to track local membership which
is causing to make these entries as stale in RP and FHR.
Fix : Made the change such that it removes the upstream reference
if it is added to track the local memberships.
lib, bgpd: Disallow processing of duplicate config.
This fix aims to reduce the load on BGPD when certain
exisiting configurations are replayed.
Specifically, the fix prevents BGPD from processing
routes when the following already existing configurations
are replayed:
1) A match criteria is configured within a route-map.
2) When "call" is invoked within a route-map.
3) When a route-map is tied to a BGP neighbor.
vishaldhingra [Thu, 9 May 2019 03:53:35 +0000 (20:53 -0700)]
lib : Child rmap not sending an update to parent rmap,if there is an update in child rmap.
Route map library creates a hash table to save the dependency binding.
route-map LRM permit 1
call rLRM
Whenever there is change in child routemap(rLRM), it tries to
find the dependency mapping with the child route map MATCH event
and it fails.The handing of match add and match delete was missing
to get the correct dependency,here it's LRM.
This fix would correct the flow to get the correct dependency.
Mark Stapp [Tue, 14 May 2019 16:21:29 +0000 (12:21 -0400)]
zebra: [7.1] remove vrf LSPs when vrf is deleted
Try to remove any LSPs associated with a vrf when the vrf is
deleted. The vrf code was calling a helpful zebra_mpls api,
but that api was basically a no-op for vrfs other than
the default.
Renato Westphal [Mon, 6 May 2019 18:57:02 +0000 (15:57 -0300)]
lib, yang: disable libyang custom user types temporarily
libyang 1.0 introduced a few changes in the user types API, and
these changes made FRR incompatible with libyang 1.x. In order to
ease our migration from libyang 0.x to libyang 1.x, let's disable
our libyang custom user types temporarily so that FRR can work
with both libyang 0.x and libyang 1.x. This should be especially
helpful to the CI systems during the transition. Once the migration
to libyang 1.x is complete, this commit will be reverted.
Disabling our libyang custom user types should have only
minimal performance implications when processing configuration
transactions. The user types infrastructure should be more important
in the future to perform canonization of YANG data values when
necessary.
Donald Sharp [Sat, 4 May 2019 00:54:20 +0000 (20:54 -0400)]
doc, zebra: Remove "table X" command
This command is broken and has been broken since the introduction
of vrf's. Since no-one has complained it is safe to assume that
there is no call for this specialized linux command. Remove
from the system with extreme prejudice.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Donald Sharp [Mon, 6 May 2019 14:41:40 +0000 (10:41 -0400)]
zebra: Be consistent in how we call rib_add( and rib_delete( with tableid
The rib_add( and rib_delete( functions are there to allow
kernel interactions with the creation of routes. Fixup the
code to be consistent in the passup of the tableid.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Quentin Young [Fri, 3 May 2019 19:42:59 +0000 (19:42 +0000)]
*: use ZAPI_CALLBACK_ARGS macro for zapi handlers
This macro:
- Marks ZAPI callbacks for readability
- Standardizes argument names
- Makes it simple to add ZAPI arguments in the future
- Ensures proper types
- Looks better
- Shortens function declarations
Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>
Quentin Young [Wed, 1 May 2019 00:49:13 +0000 (00:49 +0000)]
zebra: fix zapi msg debugging dumps
When we switched to a pthread per client, we lost the ability to
correlate zapi message debugs with their handlers in zlog, because the
message was logged when it was read off the zapi socket and not right
before it was processed. Move the zapi msg hexdump to happen right
before we call the message handler.
Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>
Donald Sharp [Sat, 4 May 2019 00:09:54 +0000 (20:09 -0400)]
bgpd: Remove logically dead code assignment
The label value is set to MPLS_LABEL_NONE at the start
of the function and we never modify it, testing it for
BGP_PREVENT_VRF_2_VRF_LEAK equality will never be true
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Mark Stapp [Thu, 2 May 2019 18:43:18 +0000 (14:43 -0400)]
topotest: fix pytest deprecation warning
As of pytest 4.something, a pattern we were using in conftest.py
was deprecated. Also make a new-ish test script executable (all
the rest appear to be?)
Donald Sharp [Thu, 2 May 2019 14:37:04 +0000 (10:37 -0400)]
pimd: Stop crash in show of single interface
There exists a possiblity that we have upstream data but
at this point in time the rpf failed because there is no
path. As such the rpf interface will be NULL and we
should not necessarily trust it. Prevent a crash
Ticket: CM-24857 Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Donald Sharp [Wed, 1 May 2019 00:23:52 +0000 (20:23 -0400)]
lib: Convert table code to use new hash type
This converts the new table code to use the new hash
type provided by David.
The following test is 1 million routes installed and how
much memory we are using:
Old mem usage:
Memory statistics for zebra:
System allocator statistics:
Total heap allocated: 574 MiB
Holding block headers: 0 bytes
Used small blocks: 0 bytes
Used ordinary blocks: 536 MiB
Free small blocks: 33 MiB
Free ordinary blocks: 4600 KiB
Ordinary blocks: 0
Small blocks: 0
Holding blocks: 0
New Memory usage:
Memory statistics for zebra:
System allocator statistics:
Total heap allocated: 542 MiB
Holding block headers: 0 bytes
Used small blocks: 0 bytes
Used ordinary blocks: 506 MiB
Free small blocks: 3374 KiB
Free ordinary blocks: 33 MiB
Ordinary blocks: 0
Small blocks: 0
Holding blocks: 0
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Donald Sharp [Tue, 30 Apr 2019 23:38:15 +0000 (19:38 -0400)]
lib: Make _find functions treat the head as const
The head of a list should not change for find functions. Probably
are others that should be considered but these changes can come
in as needed I believe.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Donald Sharp [Tue, 30 Apr 2019 22:04:57 +0000 (18:04 -0400)]
zebra: Remove linked list and replace with new LIST
The `struct rib_dest_t` was being used to store the linked
list of rnh's associated with the node. This was taking up
a bunch of memory. Replace with new data structure supplied
by David and see the memory reductions associated with 1 million
routes in the zebra rib:
Old:
Memory statistics for zebra:
System allocator statistics:
Total heap allocated: 675 MiB
Holding block headers: 0 bytes
Used small blocks: 0 bytes
Used ordinary blocks: 567 MiB
Free small blocks: 39 MiB
Free ordinary blocks: 69 MiB
Ordinary blocks: 0
Small blocks: 0
Holding blocks: 0
New:
Memory statistics for zebra:
System allocator statistics:
Total heap allocated: 574 MiB
Holding block headers: 0 bytes
Used small blocks: 0 bytes
Used ordinary blocks: 536 MiB
Free small blocks: 33 MiB
Free ordinary blocks: 4600 KiB
Ordinary blocks: 0
Small blocks: 0
Holding blocks: 0
`struct rnh` was moved to rib.h because of the tangled web
of structure dependancies. This data structure is used
in numerous places so it should be ok for the moment.
Future work might be needed to do a better job of splitting
up data structures and function definitions.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
projects / mirror_frr.git / log