On 10/04/2018 11:30 PM, Vieri Di Paola via Shorewall-users wrote:
> I finally got it working, but I still have a few doubts (see below).
> 
> On Thursday, October 4, 2018, 12:25:05 PM GMT+2, Vieri Di Paola via 
> Shorewall-users <shorewall-users@lists.sourceforge.net> wrote: 
>>
>> If I were to move to my 3-ppp setup, I guess I would need to specify the ppp 
>> option "defaultroute" for each ppp interface, right?
> 
> Setting up the "defaultroute" option on all three ppp interfaces did the 
> trick.
> 
>> Also, when starting or restarting shorewall for the first time, I get this 
>> message:
>> WARNING: Interface enp7s0 is not usable -- Provider ISP1 (1) not Started
>> (same thing for the other 2 providers)
>> This also happened when I used the pppoe links instead of ethernet.
>> However, all 3 providers are up and running, ie., I can successfully ping to 
>> a remote host through their interfaces.
>> I need to manually run "shorewall enable INTERFACE" and restart shorewall. 
>> No issues from this point onwards.
>> So why is Shorewall complaining about the interfaces? How does it decide if 
>> it's "usable"?
> 
> This can still happen, but it's not always reproducible.
> 
> However, there's one more awkward situation I've noticed today.
> In my fully working 3-pppoe gateway setup, I decided to physically shut down 
> one of my 3 provider modems and boot it back up again to see what happened 
> (say, ppp3). I was unable to access Internet via ppp3 even after a long while 
> (to allow it to sync).
> My ppp options are as follows:
> 
> noauth
> defaultroute
> persist
> holdoff 0
> maxfail 0
> noaccomp noccp nobsdcomp nodeflate nopcomp novj novjccomp
> 
> 
> I still haven't found the time to fully check what went wrong, but here's 
> what I did to bring the link back up:
> 
> First, I manually restarted the ppp3 interface and waited for full sync. I 
> checked the logs and confirmed that it was all up and running (authenticated, 
> etc.).
> 
> However, I still didn't have connectivity through this pppoe link.
> 
> I then issued "shorewall reenable ppp3", and SW reported that it had first 
> "stopped" the ppp3 provider and then "started" it (which means that shorewall 
> didn't automatically disable it while the modem was rebooting).

Shorewall will NEVER automatically disable an interface when it goes
down; that is the job of a link monitor like FooLSM.
> 
> I still had no connectivity through this pppoe link.
> 
> Finally, a shorewall restart (full stop and start) actually DID solve the 
> issue. I magically got my ppp3 link working again.
> So, of course, I'm worried that if there's a power outage or if someone 
> reboots the modems then the gateway might get cut off from the Internet if 
> shorewall doesn't restart.

I can't comment without seeing the difference between the ruleset after
'reenable' and the ruleset after 'restart'.

> 
>> I have the following: 
>> # sysctl  net.core.default_qdisc
>> net.core.default_qdisc = fq_codel
>> # sysctl  net.ipv4.tcp_congestion_control
>> net.ipv4.tcp_congestion_control = cdg
>> # ip addr show | grep qdisc
>> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group 
>> default qlen 
>> 10002: enp5s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast 
>> state UP group default qlen 
>> 10003: enp6s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc tbf state UP 
>> group default qlen 
>> 10004: enp7s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc tbf state UP 
>> group default qlen 
>> 10005: enp8s5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc tbf state UP 
>> group default qlen 
>> 10006: enp10s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast 
>> state UP group default qlen 1000
>> Why do I see pfifo_fast and tbf instead of fq_codel?
> 
> OK, I think I got this one.
> I had TC_ENABLED=Simple. Disabling it activates fq_codel on all "outer 
> interfaces", but not on the "inner" eth interfaces.
> 
> # ip addr show | grep qdisc
> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group 
> default qlen 1000
> 2: enp5s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state 
> UP group default qlen 1000
> 3: enp6s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP 
> group default qlen 1000
> 4: enp7s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP 
> group default qlen 1000
> 5: enp8s5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP 
> group default qlen 1000
> 6: enp10s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state 
> UP group default qlen 1000
> 7: ppp1: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1492 qdisc fq_codel 
> state UNKNOWN group default qlen 3
> 8: ppp2: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1492 qdisc fq_codel 
> state UNKNOWN group default qlen 3
> 10: ppp3: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1492 qdisc fq_codel 
> state UNKNOWN group default qlen 3
> 
> I still don't know why some interfaces are set to use pfifo_fast, or if it's 
> recommended or not for a gateway/router.
> 

Which has nothing to do with Shorewall...

-Tom
-- 
Tom Eastep        \   Q: What do you get when you cross a mobster with
Shoreline,         \     an international standard?
Washington, USA     \ A: Someone who makes you an offer you can't
http://shorewall.org \   understand
                      \_______________________________________________

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Shorewall-users mailing list
Shorewall-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/shorewall-users

Reply via email to