[BUG] XG v16/17 PPPOE doesn't endlessly tries to reconnect

Hi guillaume bottollier 

Apologies for the inconvenience caused by this. Would it be possible to enable the support access tunnel on your appliance and PM me with the ID for further investigation?

Thanks,

Hi FloSupport 

hi have this weird things happens again...

PPPOE interface is disconnected, nothing on the log about PADI/PADO/LCP.

Just clic on the connect button and the session goes up normally without any error... Any hint on this ?!

Hi guillaume bottollier 

Thank you for providing details through PM, we will look into it and inform you further.

Please allow us some time.

Has there been any progress with this? We have some customers that still use PPPoE connections and one customer in particular has had this happen 3 or 4 times this week.

Before this I thought that the problem was having when there was a long outage on the ISP connection, like maybe XG will try for 1 hour and then give up. Now i'm not so sure this is related. There appear to be some big time jumps in the logs too.

This customer has some punctuation characters in the credentials used for the PPPoE connection ("*" and "?"), and is running HA. I can't see any other things different from customers where this problem doesn't happen.

James

Hello James,

Thank you for contacting the Sophos Community.

If possible please try creating a new post, not necessary right now.

Could you please send me by PM the output of the syslog.log and let us know the time of the issue, as well as a screenshot of the PPPoE interface configuration.

Do you have LCP failure enabled?

Regards,

hi emmosophos 

"LCP echo interval" and "LCP failure" are activated by default with default values... they won't help reconnecting in case of pppoe disconnection, when sophos fails to reconnect.

Hello,

This issue is being investigated now for GES and Development under NC-62029

Regards,

emmosophos 

Thank you for this update, i hope this issue will be quickly and seriously solved !

Hey

Add in that this issue also affects V18 MR1-396 also

hi emmosophos

just had the failure, again and again.

 Step 1 / WAN port is Disconnected whithout ANY obvious reason !

Step 2 / Just clic on the "Connect Now" without doing NOTHING else (NO modem restart, NO lack of xDSL sync...)

Step 3 / a few seconds later :

It's now URGENT we get a solution !

Hello Guillaume,

Thank you for the follow-up.

Have you created a case with Support? I don't see you have shared one. 

I do see DEV is working on this issue, but in order for me to bring your case to them I would need a Case ID, so I can escalate it.

It seems they have identified what is causing this for the last update in NC-62029

Regards,

please correct this bug asap , it's annoying a lot of customers

Thank you

please correct this bug asap , it's annoying a lot of customers

Thank you

I am getting the feeling that this has been introduced from some of the recent HotFix updates in V17.5

I have a customer with an XG230 running 17.5.MR9

For 3 months from install, they have had no issue with PPPOE reconnect.

In the last 3 weeks they have had 5 occurrences of PPPOE disconnect and failure to reconnect. Again logging into the web interface and pressing connect and the connection immediately comes back up.

Customer is not on a DSL service, it is a Fixed Wifi service which has been stable for quite some time, their previous XG125 had no issues running several versions of 17.x on the same link.

About the time that the SQL Injection hotfix was applied, and the external web access started with the additional authentication, this has begun happening.

hello GavinDaniels 

We have this issue since years...

I have exactly the same issues here in Melbourne Australia.

I have a customer who has a CR50iNG which is upgraded to run Sophos OS.

I am struggling with this issue as well.

They have a Fibre service with PPOE running over the top.  When the link goes down, sometimes you see in the log the PADI requests, and then it reconnects perfectly.

Yesterday I had to go onsite as their link went down, and just clicking on connect brought it back up again.

I played with dropping the Fibre connection, and everytime the PPOE re-established correctly.

Then 24 hours later, and after the fibre connection dropping twice last night, exactly the same issue.

The connection is down, and the only way to fix it is to go back onsite and reconnect.

This is a serious issue that Sophos needs to address.

I am seeing this issue on a CR50iNG device running the latest firmware.

SFOS 17.5.13 MR-13

This issue has only starting happening in the last month, and I suspect it could be to do with an possible hotfix.

Further, Sophos, please note that this device is a Cyberoam CR50iNG so I can't upgrade to V18, as it doesn't support the Cyberoam devices.

We need a fix to V17 firmware.

I do not have any insight of this issue right now, but coming from a DEV/Support perspective, this is a pain to troubleshoot and find the real root cause of this issue.

As most of those appliances are completely offline while this issue occurs and only one time per day, this appears, it is hard to "live debug" this issue.

So if somebody with this issue have a system, which has a "alternative" route to the internet at the same time, this would be helpful to live debug this issue. Maybe even a cellular connection to XG would be helpful, while this occurs. 

Hello LuCar,

At the present I can't help with this request, the 2 sites that are experiencing this more than once so far are critical, as in Australia we presently have a large number of work from home people, so immediate reconnection is required.

But I have the site that experienced the issue for the first time yesterday getting a secondary link. So if there is no resolution then I will be able to have them fully failed over on the backup link for immediate testing.

I will advise.

But some credence needs to be applied that the issue appears to be introduced in a hotfix. As I explained, I have 2 sites with in excess of 100 days uptime, and no firmware changes. And they have only begun exhibiting the issue in the past 20 days. And given that 1 site has experienced the issue 4 times in that 20 days, and not at all in the previous 80 gives this indication.

Regards

Gavin