OpenBGP lost session

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

OpenBGP lost session

bernd-34
Hi list,

I've got two OpenBSD (5.1-STABLE, amd64) machines running OpenBGPd.

Both of them are connected to two upstream providers each, furthermore
there are (older) Ciscos, also connecteed to the same (!) upstream
routers.

Recently, both OpenBSD machines lost their BGP session to one of the
upstream providers. On both machines the same upstream router was
affected.

Logs show this:

Sep 17 17:25:35 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
sending notification: HoldTimer expired, unknown subcode 0
Sep 17 17:25:35 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
state change Established -> Idle, reason: HoldTimer expired
Sep 17 17:25:43 hostname ospfd[5366]: desync; scheduling fib reload
Sep 17 17:25:43 hostname ospfd[5366]: reloading interface list and
routing table
Sep 17 17:25:48 hostname bgpd[15513]: nexthop 12.23.34.45 now valid:
directly connected
Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
state change Idle -> Connect, reason: Start
Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
state change Connect -> OpenSent, reason: Connection opened
Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
state change OpenSent -> OpenConfirm, reason: OPEN message received
Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
state change OpenConfirm -> Established, reason: KEEPALIVE message
received
Sep 17 17:26:20 hostname bgpd[15513]: nexthop 12.23.34.45 now valid:
directly connected

The Ciscos didn't see anything like this, their sessions didn't drop.

Any clue what was going on?

Thanks,

Bernd

Reply | Threaded
Open this post in threaded view
|

Re: OpenBGP lost session

Stuart Henderson
On 2012-09-18, Bernd <[hidden email]> wrote:

> Hi list,
>
> I've got two OpenBSD (5.1-STABLE, amd64) machines running OpenBGPd.
>
> Both of them are connected to two upstream providers each, furthermore
> there are (older) Ciscos, also connecteed to the same (!) upstream
> routers.
>
> Recently, both OpenBSD machines lost their BGP session to one of the
> upstream providers. On both machines the same upstream router was
> affected.
>
> Logs show this:
>
> Sep 17 17:25:35 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
> sending notification: HoldTimer expired, unknown subcode 0
> Sep 17 17:25:35 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
> state change Established -> Idle, reason: HoldTimer expired
> Sep 17 17:25:43 hostname ospfd[5366]: desync; scheduling fib reload
> Sep 17 17:25:43 hostname ospfd[5366]: reloading interface list and
> routing table
> Sep 17 17:25:48 hostname bgpd[15513]: nexthop 12.23.34.45 now valid:
> directly connected
> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
> state change Idle -> Connect, reason: Start
> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
> state change Connect -> OpenSent, reason: Connection opened
> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
> state change OpenSent -> OpenConfirm, reason: OPEN message received
> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45 (Upstream1):
> state change OpenConfirm -> Established, reason: KEEPALIVE message
> received
> Sep 17 17:26:20 hostname bgpd[15513]: nexthop 12.23.34.45 now valid:
> directly connected
>
> The Ciscos didn't see anything like this, their sessions didn't drop.
>
> Any clue what was going on?
>
> Thanks,
>
> Bernd
>
>

Can't tell from this. Are you running the same hold times on your openbgp
boxes as your ciscos?

Reply | Threaded
Open this post in threaded view
|

Re: OpenBGP lost session

bernd-34
Am 2012-09-18 16:34, schrieb Stuart Henderson:

> On 2012-09-18, Bernd <[hidden email]> wrote:
>> Hi list,
>>
>> I've got two OpenBSD (5.1-STABLE, amd64) machines running OpenBGPd.
>>
>> Both of them are connected to two upstream providers each,
>> furthermore
>> there are (older) Ciscos, also connecteed to the same (!) upstream
>> routers.
>>
>> Recently, both OpenBSD machines lost their BGP session to one of the
>> upstream providers. On both machines the same upstream router was
>> affected.
>>
>> Logs show this:
>>
>> Sep 17 17:25:35 hostname bgpd[1638]: neighbor 12.23.34.45
>> (Upstream1):
>> sending notification: HoldTimer expired, unknown subcode 0
>> Sep 17 17:25:35 hostname bgpd[1638]: neighbor 12.23.34.45
>> (Upstream1):
>> state change Established -> Idle, reason: HoldTimer expired
>> Sep 17 17:25:43 hostname ospfd[5366]: desync; scheduling fib reload
>> Sep 17 17:25:43 hostname ospfd[5366]: reloading interface list and
>> routing table
>> Sep 17 17:25:48 hostname bgpd[15513]: nexthop 12.23.34.45 now valid:
>> directly connected
>> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45
>> (Upstream1):
>> state change Idle -> Connect, reason: Start
>> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45
>> (Upstream1):
>> state change Connect -> OpenSent, reason: Connection opened
>> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45
>> (Upstream1):
>> state change OpenSent -> OpenConfirm, reason: OPEN message received
>> Sep 17 17:26:05 hostname bgpd[1638]: neighbor 12.23.34.45
>> (Upstream1):
>> state change OpenConfirm -> Established, reason: KEEPALIVE message
>> received
>> Sep 17 17:26:20 hostname bgpd[15513]: nexthop 12.23.34.45 now valid:
>> directly connected
>>
>> The Ciscos didn't see anything like this, their sessions didn't
>> drop.
>>
>> Any clue what was going on?
>>
>> Thanks,
>>
>> Bernd
>>
>>
>
> Can't tell from this. Are you running the same hold times on your
> openbgp
> boxes as your ciscos?

Hi, yes, it's 90 sec on the Ciscos as well as for BGPd (default is 90
sec).

Best,

Bernd