Age | Commit message (Collapse) | Author |
|
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
just move some stuff
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
|
|
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
|
|
|
|
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
$ cd $GOBGP/gobgp/lib
$ go version
go version go1.5 linux/amd64
$ go build -buildmode=c-shared -o libgobgp.so *.go
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Signed-off-by: Hiroshi Yokoi <yokoi.hiroshi@po.ntts.co.jp>
|
|
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
|
|
--cpus option can be used to specify to the number of CPUs to be
used. IOW, the specified number is passed to runtime.GOMAXPROCS().
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
if you don't specify ttl, then we use the OS-default ttl.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
distribute-policy is deprecated name.
We chose to use the name in-policy.
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
non-route-server-client peers don't have local rib.
Reported-by: Pavel Odintsov <pavel.odintsov@gmail.com>
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
gobgp asks for non existing thing (e.g. local rib for non route server
peer), gobgpd doesn't send anything so gobgp will hang up.
This fixes gobgpd to send an empty response.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
make it more pretty when marshaling flowspec nlri
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Signed-off-by: Hiroshi Yokoi <yokoi.hiroshi@po.ntts.co.jp>
|
|
Signed-off-by: Hiroshi Yokoi <yokoi.hiroshi@po.ntts.co.jp>
|
|
Signed-off-by: Hiroshi Yokoi <yokoi.hiroshi@po.ntts.co.jp>
|
|
You can enable the feature like the following:
[Neighbors]
[[Neighbors.NeighborList]]
[Neighbors.NeighborList.NeighborConfig]
PeerAs = 65001
NeighborAddress = "10.0.255.1"
[Neighbors.NeighborList.EbgpMultihop]
[Neighbors.NeighborList.EbgpMultihop.EbgpMultihopConfig]
Enabled = true
MultihopTtl = 8
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
scenario_test is also added
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Actually comapring all the attributes to see if packing NLRIs. It
takes long but gobgpd sends minimum messages.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
This reverts commit 51494759aded7098e151869e66dd5b2c4d96ecfd.
gobgpd still sends too more updates rather than Quagga. Revert this.
|
|
Currently, mrt inject wrongly use gobgpd's AS number and
RouterID. Instead use these in the dump file.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Make work faster if possible (with consuming more memory). The number
is near the full destinations.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
mrt dump file should include nexthop attribute
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
The draft was promoted to IETF WG item.
It is already enqueued to RFC editors in Aug. 2015. It's expected to be published as RFC soon.
Signed-off-by: Jun-ya Kato <kato.junya@lab.ntt.co.jp>
|
|
Commiting helper function on MRTHeader to return a golang time.Time
from the timestamp. Also commiting the relevant test function.
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
we can't parse Path.nlri field (whose type is []byte) without route family.
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Without hurting the performance (the latency) much, GetBestPathList()
tries to sort Paths in the order that many routes could be packed into
one update message.
I did an experiment that the full routes are pushed to gobgpd (via
mrt), then another peer connects to gobgpd.
Without this patch, gobgpd sends 550,958 update messages to another
peer.
With this patch, gobgpd sends 250,606 update messages.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
With experiments about the full routes, we hit the limit. Let's remove
it.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
Not tested on Mac OS X but at least buildable.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Use more appropriate names
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Signed-off-by: ISHIDA Wataru <ishida.wataru@lab.ntt.co.jp>
|
|
With this patch, bobgpd packs multiple NLRIs into one BGP update
message (IOW, multiple paths into one update message).
We do only with ipv4 since we could have lots of routes. If you don't
have lots, it's not worth having the complicated logic for such route
families.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|
|
Seems that no good reason to have a separate repo for gozebra.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
|