Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-23 Thread Julien Goodwin
On 23/12/10 21:34, Florian Weimer wrote:
 * Julien Goodwin:
 
 For my SRX at the office back when I installed it (9.6 IIRC) *TCP*
 keepalives would not extend session timeouts, but *SSH* keepalives
 worked very well, that's the ServerAliveInterval setting in OpenSSH.
 
 Typically, TCP keepalives happen at such long intervals that they do
 not keep firewall state alive.

In my specific case (whinging admin in internal IT, not production) they
were at least every minute.

We do actually have some systems that are so old/weird they don't
support the ServerAliveInterval, but they're all fairly minor so it's
not usually a problem.

-- 
Julien Goodwin
Studio442
Blue Sky Solutioneering
___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp


Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-23 Thread Florian Weimer
* Julien Goodwin:

 For my SRX at the office back when I installed it (9.6 IIRC) *TCP*
 keepalives would not extend session timeouts, but *SSH* keepalives
 worked very well, that's the ServerAliveInterval setting in OpenSSH.

Typically, TCP keepalives happen at such long intervals that they do
not keep firewall state alive.

-- 
Florian Weimerfwei...@bfk.de
BFK edv-consulting GmbH   http://www.bfk.de/
Kriegsstraße 100  tel: +49-721-96201-1
D-76133 Karlsruhe fax: +49-721-96201-99

___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp


Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-22 Thread Maciej Jan Broniarz


Dnia 10-12-20 18:04 użytkownik „Mark Kamichoff” p...@prolixium.com
napisał:

On Mon, Dec 20, 2010 at 10:18:27AM -0600, Chris Adams wrote:
 I don't know about the SRX, but I know with the SSG, the ScreenOS
 default timeout for TCP sessions was way too low (IIRC something like
 5 minutes) and would cause that.  I turned on SSH keepalives to avoid
 the timeout.

Yep, the SRX does the same thing with regards to timeouts.  The timeout
is 30 minutes for SSH by default, but you can extend it to longer by
adding a custom inactivity-timeout to the junos-ssh application:

{primary:node0}
p...@orb show configuration applications
application junos-ssh inactivity-timeout 3600;

Does junos-ssh applies to any ssh traffic - the one to the srx itself, and
the one to the servers behind an SRX firewall?

Alternatively, you can set the tcp-rst option on the appropriate
zone(s), which will cause SSH sessions to disconnect immediately when
data is sent over an SSH session that's timed-out already:

{primary:node0}[edit]
p...@orb# show security zones security-zone trust
tcp-rst;
[...]


I'll try that.

Thanks :)

Mjb



___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp

Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-22 Thread Mark Kamichoff
On Wed, Dec 22, 2010 at 07:43:30PM +0100, Maciej Jan Broniarz wrote:
 {primary:node0}
 p...@orb show configuration applications
 application junos-ssh inactivity-timeout 3600;
 
 Does junos-ssh applies to any ssh traffic - the one to the srx itself,
 and the one to the servers behind an SRX firewall?

In my experience, both.

(unless you're connected via the fxp0 interface in a cluster, which I
believe is excluded from the flow/state tracking)

- Mark

-- 
Mark Kamichoff
p...@prolixium.com
http://www.prolixium.com/


signature.asc
Description: Digital signature
___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp

Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-21 Thread Alfred Schweder
Hello

  I don't know about the SRX, but I know with the SSG, the ScreenOS
  default timeout for TCP sessions was way too low (IIRC something like
  5 minutes) and would cause that.  I turned on SSH keepalives to avoid
  the timeout.

Does SRX support ssh keepalive (like M- or J-serie)?

SSGs drop the ssh session if they get a keepalive.
Juniper isn't willing to fix this in the near future ;-(

Regards, Alf

___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp


Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-21 Thread Julien Goodwin
On 22/12/10 04:53, Alfred Schweder wrote:
 Does SRX support ssh keepalive (like M- or J-serie)?
 
 SSGs drop the ssh session if they get a keepalive.
 Juniper isn't willing to fix this in the near future ;-(

For my SRX at the office back when I installed it (9.6 IIRC) *TCP*
keepalives would not extend session timeouts, but *SSH* keepalives
worked very well, that's the ServerAliveInterval setting in OpenSSH.

It's arguable whether this is correct, but as everything I use supports
the SSH keepalives it works for me, as well as being much better at
noticing when a connection *does* drop.


-- 
Julien Goodwin
Studio442
Blue Sky Solutioneering
___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp


Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-21 Thread Barny Sanchez
Yes it is supported in SRX, and you are right not supported in ScreenOS, it 
drops it.

Thanks!

- Barny

On Dec 21, 2010, at 8:08 PM, Julien Goodwin jgood...@studio442.com.au wrote:

 On 22/12/10 04:53, Alfred Schweder wrote:
 Does SRX support ssh keepalive (like M- or J-serie)?
 
 SSGs drop the ssh session if they get a keepalive.
 Juniper isn't willing to fix this in the near future ;-(
 
 For my SRX at the office back when I installed it (9.6 IIRC) *TCP*
 keepalives would not extend session timeouts, but *SSH* keepalives
 worked very well, that's the ServerAliveInterval setting in OpenSSH.
 
 It's arguable whether this is correct, but as everything I use supports
 the SSH keepalives it works for me, as well as being much better at
 noticing when a connection *does* drop.
 
 
 -- 
 Julien Goodwin
 Studio442
 Blue Sky Solutioneering
 ___
 juniper-nsp mailing list juniper-nsp@puck.nether.net
 https://puck.nether.net/mailman/listinfo/juniper-nsp

___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp


Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-20 Thread Chris Adams
Once upon a time, Maciej Jan Broniarz gau...@gausus.net said:
 I have a bunch of servers connected to an Juniper SRX. When I log on them via 
 ssh everything is fine, but when I leave the session idle for a few minutes 
 it freezes and i have to close the connection and login again. What might be 
 the issue here?
 The problem only occurs when I connect to a server behind an SRX device.

I don't know about the SRX, but I know with the SSG, the ScreenOS
default timeout for TCP sessions was way too low (IIRC something like 5
minutes) and would cause that.  I turned on SSH keepalives to avoid the
timeout.

-- 
Chris Adams cmad...@hiwaay.net
Systems and Network Administrator - HiWAAY Internet Services
I don't speak for anybody but myself - that's enough trouble.
___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp


Re: [j-nsp] Juniper SRX and ssh freeze

2010-12-20 Thread Mark Kamichoff
On Mon, Dec 20, 2010 at 10:18:27AM -0600, Chris Adams wrote:
 I don't know about the SRX, but I know with the SSG, the ScreenOS
 default timeout for TCP sessions was way too low (IIRC something like
 5 minutes) and would cause that.  I turned on SSH keepalives to avoid
 the timeout.

Yep, the SRX does the same thing with regards to timeouts.  The timeout
is 30 minutes for SSH by default, but you can extend it to longer by
adding a custom inactivity-timeout to the junos-ssh application:

{primary:node0}
p...@orb show configuration applications 
application junos-ssh inactivity-timeout 3600;

The above configuration increases the inactivity timeout to an hour.
For me, I had one session built before I made that change, and one after
(look at the timeout value):

{primary:node0}
p...@orb show security flow session destination-prefix 10.3.8.18/32 node 0 
node0:
--

Session ID: 8824, Policy name: inbound/4, State: Active, Timeout: 1796, Valid
  In: 10.3.7.149/63197 -- 10.3.8.18/22;tcp, If: reth0.0, Pkts: 61, Bytes: 6901
  Out: 10.3.8.18/22 -- 10.3.7.149/63197;tcp, If: reth2.0, Pkts: 37, Bytes: 9556

Session ID: 8832, Policy name: inbound/4, State: Active, Timeout: 3594, Valid
  In: 10.3.7.149/63198 -- 10.3.8.18/22;tcp, If: reth0.0, Pkts: 55, Bytes: 6445
  Out: 10.3.8.18/22 -- 10.3.7.149/63198;tcp, If: reth2.0, Pkts: 34, Bytes: 7288
Total sessions: 2

Alternatively, you can set the tcp-rst option on the appropriate
zone(s), which will cause SSH sessions to disconnect immediately when
data is sent over an SSH session that's timed-out already:

{primary:node0}[edit]
p...@orb# show security zones security-zone trust   
tcp-rst;
[...]

Hope this helps!

- Mark

-- 
Mark Kamichoff
p...@prolixium.com
http://www.prolixium.com/


signature.asc
Description: Digital signature
___
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp