Free unhashable (duplicate NHEs from the kernel) via ID table
cleanup. Since the NHE ID hash table contains extra entries,
that's the one we need to be calling zebra_nhg_hash_free()
on, otherwise we will never free the unhashable NHEs.
This was found via a memleak:
==1478713== HEAP SUMMARY:
==1478713== in use at exit: 10,267 bytes in 46 blocks
==1478713== total heap usage: 76,810 allocs, 76,764 frees, 3,901,237 bytes allocated
==1478713==
==1478713== 208 (88 direct, 120 indirect) bytes in 1 blocks are definitely lost in loss record 35 of 41
==1478713== at 0x483BB1A: calloc (vg_replace_malloc.c:762)
==1478713== by 0x48E35E8: qcalloc (memory.c:110)
==1478713== by 0x451CCB: zebra_nhg_alloc (zebra_nhg.c:369)
==1478713== by 0x453DE3: zebra_nhg_copy (zebra_nhg.c:379)
==1478713== by 0x452670: nhg_ctx_process_new (zebra_nhg.c:1143)
==1478713== by 0x4523A8: nhg_ctx_process (zebra_nhg.c:1234)
==1478713== by 0x452A2D: zebra_nhg_kernel_find (zebra_nhg.c:1294)
==1478713== by 0x4326E0: netlink_nexthop_change (rt_netlink.c:2433)
==1478713== by 0x427320: netlink_parse_info (kernel_netlink.c:945)
==1478713== by 0x432DAD: netlink_nexthop_read (rt_netlink.c:2488)
==1478713== by 0x41B600: interface_list (if_netlink.c:1486)
==1478713== by 0x457275: zebra_ns_enable (zebra_ns.c:127)
Repro with:
ip next add id 1 blackhole
ip next add id 2 blackhole
valgrind /usr/lib/frr/zebra
Signed-off-by: Stephen Worley <sworley@cumulusnetworks.com>
The rtadv code has two types of sockets:
a) namespace -> Where each zvrf get's it's own socket
b) vrf lite -> Where we get 1 socket for everything
When we were terminating a vrf we were *always*
killing the (b) socket. This is a mistake in
that other vrf's may need to be communicating.
Modify the code on vrf shutdown to only disable
that vrf's event processing and when we actually
terminate we shut the socket.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
This macro is undefined if vnc is disabled, and while it defaults to 0,
this is still wrong and causes issues with -Werror
Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>
There is configuration in LDP to only create labels for
host-routes. If the user remove this configuration the code
was not readvertising non-host routes to it's LDP neighbors.
The issue is the same in reverse also. If the user adds this
configuration on an active LDP session the non-host routes were
not withdrawn.
Signed-off-by: Lynne Morrison <lynne@voltanet.io>
We are seeing this crash:
New LWP 7673]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Core was generated by `/usr/lib/frr/ospf6d -d -F datacenter -M snmp -A ::1'.
Program terminated with signal SIGABRT, Aborted.
(gdb) bt
vtysh=vtysh@entry=0) at lib/command.c:1288
(gdb)
The command entered is `debug ospf6 lsa inter-router examin`. Code
inspection leads us to the fact that FRR is declaring the data as
const but we are attempting to modify it, causing the crash.
Remvoe the const of this set/get and let things work.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
hook_register() invocations generally are in some initialization
function and not looped over or similar. We can use a static struct
hookent variable for the hook list entry in 99.999% of cases, so let's
do that and not malloc() memory.
Signed-off-by: David Lamparter <equinox@diac24.net>
This is most of the old code bolted on top of the new "backend"
infrastructure. It just wraps around zlog_fd() with the string search.
Originally-by: Stephen Worley <sworley@cumulusnetworks.com>
Signed-off-by: David Lamparter <equinox@diac24.net>
In some cases we really don't want to clean up things even when exiting
(i.e. to keep the logging subsystem going.) This adds a flag on MGROUPs
to indicate that.
[v2: add "(active at exit)" marker text to debug memstats-at-exit]
Signed-off-by: David Lamparter <equinox@diac24.net>
This is a full rewrite of the "back end" logging code. It now uses a
lock-free list to iterate over logging targets, and the targets
themselves are as lock-free as possible. (syslog() may have a hidden
internal mutex in the C library; the file/fd targets use a single
write() call which should ensure atomicity kernel-side.)
Note that some functionality is lost in this patch:
- Solaris printstack() backtraces are ditched (unlikely to come back)
- the `log-filter` machinery is gone (re-added in followup commit)
- `terminal monitor` is temporarily stubbed out. The old code had a
race condition with VTYs going away. It'll likely come back rewritten
and with vtysh support.
- The `zebra_ext_log` hook is gone. Instead, it's now much easier to
add a "proper" logging target.
v2: TLS buffer to get some actual performance
Signed-off-by: David Lamparter <equinox@diac24.net>
We are still seeing cbit test failures in the ci system. I am
gonna try extending the timeout a bit more as that 8 seconds
doesn't seem to be long enough.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
- each statistics is encapsulated into concatenated "<afi><safi>" value.
- the json encoding for floating and double values is using json api
double api. this change is done for bgp statistics.
- the lines over 80 characters have been handled.
Signed-off-by: Philippe Guibert <philippe.guibert@6wind.com>
this command is a shortcut to facilitate the extraction of statistics
for all afi/safi related to one bgp instance.
the command is: show bgp [vrf XX] statistics-all [json]
Signed-off-by: Philippe Guibert <philippe.guibert@6wind.com>
safis that use a route distinguisher in bgp tables, and as such
introduce a two level hierarchy on the bgp table, must be made available
to statistics too.
Signed-off-by: Philippe Guibert <philippe.guibert@6wind.com>
add json support for show bgp statistics command.
The title of the stats entry is aggregated without spaces.
Signed-off-by: Philippe Guibert <philippe.guibert@6wind.com>
Implement tests to verify BGP link-bandwidth and weighted ECMP
functionality. These tests validate one of the primary use cases for
weighted ECMP (a.k.a. Unequal cost multipath) using BGP link-bandwidth:
https://tools.ietf.org/html/draft-mohanty-bess-ebgp-dmz
The included tests are:
Test #1: Test BGP link-bandwidth advertisement based on number of multipaths
Test #2: Test cumulative link-bandwidth propagation
Test #3: Test weighted ECMP - multipath with next hop weights
Test #4: Test weighted ECMP rebalancing upon change (link flap)
Test #5: Test weighted ECMP for a second anycast IP
Test #6: Test paths with and without link-bandwidth - receiver should resort to regular ECMP
Test #7: Test different options for processing link-bandwidth on the receiver
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
The BGP Router MAC extended community should be unique and not occur
multiple times. In a VRF-to-VRF route-leak scenario where EVPN routes
from a source VRF are leaked into the target VRF and then injected
back into EVPN from the target VRF, the resulting route had more than
one RMAC. With this fix, the resulting route will have only the
target VRF's RMAC.
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
The EVPN advertise route-map may generate extended communities for an IPv4
or IPv6 route injected into EVPN as type-5. If so, allow for it and add
to it.
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
Reviewed-by: Don Slice <dslice@cumulusnetworks.com>
Reviewed-by: Donald Sharp <sharpd@cumulusnetworks.com>
Support configurable options to control how link bandwidth is handled
by the receiver. The default behavior is to automatically honor the
link bandwidths received and use it to perform a weighted ECMP BUT only
if all paths in the multipath have associated link bandwidth; if one or
more paths do not have link bandwidth, normal ECMP is performed among
the multipaths. This behavior is as recommended by
https://tools.ietf.org/html/draft-ietf-idr-link-bandwidth.
The additional options available are to (a) completely ignore any link
bandwidth (i.e., weighted ECMP is effectively disabled), (b) skip paths
in the multipath which do not have link bandwidth and perform weighted
ECMP among the other paths (if at least some paths have the bandwidth)
or (c) use a default weight (value chosen is 1) for the paths which
do not have link bandwidth.
The command syntax is
bgp bestpath bandwidth <ignore|skip-missing|default-weight-for-missing>
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
When announcing ourselves as the next hop (e.g., to EBGP peers), if the
best path has the link bandwidth extended community and it is transitive,
change the value of the link bandwidth to the cumulative downstream
bandwidth (sum of the link bandwidths of all our multipaths) as this
makes the most sense. It is also implied by
https://tools.ietf.org/html/draft-mohanty-bess-ebgp-dmz. Of course, do
not override the link bandwidth if it has been specified by policy.
Note: Transitive extended communities will be automatically passed along
to EBGP peers; this commit is updating the value that is announced to
something that is the most appropriate.
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
Reviewed-by: Donald Sharp <sharpd@cumulusnetworks.com>
Reviewed-by: Don Slice <dslice@cumulusnetworks.com>
Implement the code to handle the other route-map options to generate
the link bandwidth, namely, to use the cumulative bandwidth or to
base this on the number of multipaths. In the latter case, a reference
bandwidth is internally chosen - the implementation uses a value of
1 Gbps.
These additional options mean that the prefix may need to be advertised
if there is a link bandwidth change, which is a new criteria. Define a
new path (change) flag to support this and implement the advertisement.
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
Reviewed-by: Donald Sharp <sharpd@cumulusnetworks.com>
Reviewed-by: Don Slice <dslice@cumulusnetworks.com>
The BGP link bandwidth extended community must not be repeated. If the
attribute already carries this and the route-map specifies a new value,
the implementation will honor the policy configuration and overwrite
the existing values.
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
Certain extended communities cannot be repeated. An example is the
BGP link bandwidth extended community. Enhance the extended community
add function to ensure uniqueness, if requested.
Note: This commit does not change the lack of uniqueness for any of
the already-supported extended communities. Many of them such as the
BGP route target can obviously be present multiple times. Others like
the Router's MAC should most probably be present only once. The portions
of the code which add these may already be structured such that duplicates
do not arise.
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
Perform weighted ECMP if the multipaths have link bandwidth. This involves
assigning weights to each of the next hops associated with the prefix based
on the link bandwidth of the corresponding path as a factor of the total
(cumulative) link bandwidth for the prefix. The weight values used are
between 1 and 100. Weights are assigned only if all paths in the multipath
have link bandwidth, otherwise any bandwidths are ignored and regular
ECMP is performed. This is as recommended in
https://tools.ietf.org/html/draft-ietf-idr-link-bandwidth
A subsequent commit will implement additional (user-configurable) behaviors.
Signed-off-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
Reviewed-by: Donald Sharp <sharpd@cumulusnetworks.com>
Reviewed-by: Don Slice <dslice@cumulusnetworks.com>
