Hey David,
To follow-up on an off-list discussion we had on GR, here's that tweak.
I gather you're also concerned about internal queueing and batching to
minimise CPU churn.
This patch set doesn't co-ordinate tracking of EoR-received state across
all peers, but the current code doesn't have that either, and this set
shouldn't of itself be an impediment to that.
commit 97dc9be5a0ae9ec3b3212d566ba56a39d2dae17e
Author: Paul Jakma <[email protected]>
Date: Fri May 8 22:03:48 2015 +0100
bgpd: Stateful GR R-bit flag should only apply to boot-up peers
As noted by David Lamparter, the stateful R-bit flag, to ensure R-bit was
set for the first session to peers, also would apply to peers added later.
Which just doesn't make sense - having added a peer means you havn't
restarted per se. Limit it to peers created from the config file.
Note, there's still a slight race in that a peer could be configured
immediately after start-up, but before the boot-peers have synced up, but
hey. Rare case. To fix that, the R-bit needs to become dependent on an
additional bit of global state: whether all EoRs have been received yet from
non-restarted peers.
* bgpd.h: (struct bgp_master), add a global flag to indicate when the
config file has been read.
* bgp_main.c: (main) Set it after vty_read_config is done.
* bgpd.c: (peer_new) Only set the state flag to set the GR R-bit for peers
that are being created while reading the config file.
diff --git a/bgpd/bgp_main.c b/bgpd/bgp_main.c
index 5026b5e..70f5ce8 100644
--- a/bgpd/bgp_main.c
+++ b/bgpd/bgp_main.c
@@ -433,7 +433,8 @@ main (int argc, char **argv)
/* Parse config file. */
vty_read_config (config_file, config_default);
-
+ bm->config_file_done = true;
+
/* Start execution only if not in dry-run mode */
if(dryrun)
return(0);
diff --git a/bgpd/bgpd.c b/bgpd/bgpd.c
index 5c2d880..f515137 100644
--- a/bgpd/bgpd.c
+++ b/bgpd/bgpd.c
@@ -827,8 +827,11 @@ peer_new (struct bgp *bgp)
peer->orf_plist[afi][safi] = NULL;
}
SET_FLAG (peer->sflags, PEER_STATUS_CAPABILITY_OPEN);
- SET_FLAG (peer->sflags, PEER_STATUS_GR_SEND_R_BIT);
-
+
+ /* Interactively configured peers shouldn't be told we've restarted */
+ if (!bm->config_file_done)
+ SET_FLAG (peer->sflags, PEER_STATUS_GR_SEND_R_BIT);
+
/* Create buffers. */
peer->ibuf = stream_new (BGP_MAX_PACKET_SIZE);
peer->obuf = stream_fifo_new ();
@@ -5396,6 +5399,7 @@ bgp_master_init (void)
bm->port = BGP_PORT_DEFAULT;
bm->master = thread_master_create ();
bm->start_time = bgp_clock ();
+ bm->config_file_done = false;
}
diff --git a/bgpd/bgpd.h b/bgpd/bgpd.h
index 8cfb81b..3e855d4 100644
--- a/bgpd/bgpd.h
+++ b/bgpd/bgpd.h
@@ -60,6 +60,8 @@ struct bgp_master
#define BGP_OPT_MULTIPLE_INSTANCE (1 << 1)
#define BGP_OPT_CONFIG_CISCO (1 << 2)
#define BGP_OPT_NO_LISTEN (1 << 3)
+
+ bool config_file_done;
};
/* BGP instance structure. */
--
Paul Jakma [email protected] @pjakma Key ID: 64A2FF6A
Fortune:
disks spinning backwards - toggle the hemisphere jumper.
_______________________________________________
Quagga-dev mailing list
[email protected]
https://lists.quagga.net/mailman/listinfo/quagga-dev
