Hi, Mikrotik Users
On my PPPOE server (ROS 3.13) continues PPPOE DIE,
So PPPOE connected, is working very well, but after some minutes it dies, no traffic is passing through it.
Such problem there was with PPPT on earlier versions, but on 3.13 this problem we can see just with PPPOE,


When this problem will be fixed?

Thanks With Regards
Karapet Aznavuryan

as i say before , this problem was exist even in 2.9 maybe you did not understand but after a lot of conversations with support and forum the answer is NOTHING.

after it die, create sup-out file and send it to support

I have already sent supout file to mikrotik team, but there is no news yet,
I hope they'll fix that problem, as it is very important for Mikrotik Users.
Almost every Provider uses PPTP or PPPOE Servers to provide their services.
So we need it to work properly.


With Regards
Karapet Aznavuryan

i have same problem. When pppoe client request a big number of connections, pppoe freeze, but it not disconnect. Version is 3.13 ... problem solved with downgrading?

it`s curious, when clients disconnect and reconnect, connectiviti problem was solved.

sorry for my bery bery bad english

I have this problem since first version of 2.9 but mikrotik team does not care about it . even one reply !!!
maybe it is a core FEATURE !! which makes it differ from other pppoe servers.

Hello people,
probably I have a problem similar to you.
After a while my PPPoE clients connections went down and then reconnect a lot of times.
In the log I can see PEER IS NOT RESPONDING message, wireless connection is ok ( tx/rx ccq 60/95 ) with no disconnections at all;
it seems that Mtik is loosing client replies; this problem is with all types of clients ( NS2, Straightcore etc. )
RB333 - 333Mhz - Ros 3.13, I am going to change wireless card (R52H) with a new one, second step is to change RB333 with RB433.
I already sent supout file to Mikrotik, they requested also a log at time of error, no solutions till now.
Regards
Alessandro

Actually the physical link is not the problem . i have tested it at the client side at the time of the problem ( time out ) . exactly when the pppoe dies and i got time out ( client side ) i could open winbox with the MAC not with ip ! and after some seconds of working with winbox over MAC the ping of the pppoe came back !!!

so client at the pppoe die time ( time out ) have 3 ways to bypass the time out
1. disconnect and again connect
2. open winbox with MAC or some other kind of MAC layer connection
3. wait some minutes to problem goes automatically

other friends which have this problem please try the second way and say result. if there is another way please say here.

sometimes this problem occurs over 100 times an hour and sometimes once a day

Hello,
I think should be fine if Mikrotik team inform us on progress of analysis of this problem.
It is urgent and we need to be informed if a solution will be released soon.
Thanks
Alessandro

Hello,
I think should be fine if Mikrotik team inform us on progress of analysis of this problem.
It is urgent and we need to be informed if a solution will be released soon.
Thanks
Alessandro

I agree .

Estimated
As I could observe in my situation, it is not a MRU/MTU's/MSS problem. Simply it is given occasionally, and I am starting suspecting in clients that they have a level of poor sign. The solutions that encontre: To enter for cap two with winbox or neighbor by means of mac-telnet To disconnect, and reconnect. In the terminus, to do a ping to some random page in order that routeros this obliged one to resolve. I did a script in order that every 5 seconds realizae a ping to http://www.google.com, nevertheless the problem is not solved. I emphasize that the pppoe is still active, there is not disconnected, only that it freezes. I am suspecting that is a problem of loss of routes, another thing does not happen to me.
Regards

compare your setups - do you have big PPPoE server with lots of users or small router over wireless - those are completely two different things. Do you use RADIUS, queues, MRRU??

Have you enabled pppoe,debug logging to remote place to actually see what was the last thing your pppoe concentrator did?

At this point we see only stand alone - independent incidents and in this case problem might be anywhere (pppoe clients, nework structure, wireless problems)

As soon as we will be able to find something common in several setups we will have ability to start fixing it

i have sent many supout till now . not enough ?!

i have strong x86 system and i tested with some other systems.
the routeros is level 6
i have external radius server and less than 30 queues
no mrru
logged pppoe,debug to external syslog but nothing useful
no wireless at all . no link problem also . i have tested with a lot of physical link types. and i have this problem since early versions of 2.9

actually i have a lot of research to find where this problem comes from but no success.

Hello, I think that a "common" problem to all of us is that PPPoE clients disconnect, we have radius, no MRRU no wireless disconnections and supout files already sent to you. please Normis a question to you: have you tried to read our supout? Mine is not a polemic question but my impression, and probably also for other peole here, is that you ( Mikrotik support ) have not yet taken this problem into appropriate level of urgency.
I know, probably this is not Mikrotik problem, but we need to you a little bit more help than just " this is not mtik problem".
I use your products and will continue to use them but sometimes support should jump into our field to see real customers
With best regards
Alessandro

compare your setups - do you have big PPPoE server with lots of users or small router over wireless - those are completely two different things. Do you use RADIUS, queues, MRRU??

Have you enabled pppoe,debug logging to remote place to actually see what was the last thing your pppoe concentrator did?

At this point we see only stand alone - independent incidents and in this case problem might be anywhere (pppoe clients, nework structure, wireless problems)

As soon as we will be able to find something common in several setups we will have ability to start fixing it

Ive heard rumors somewhere that MT team would replace the current kernel and at this point I started to believe that it will not happens until a major version number change, perhaps. BUT I do believe that MT team knows that this freeze problem is something they need to rely on kernel developers to solve. Could this be something common? (unregistered_netdevice: ... bla bla bla thing) That is what makes me feel a bit of calm facing the situation.

Now reading your last words I may change my mind. Didn't said anything until now because MT was looking about that random queue missing thread. Anyway, I would like to help this out. What can I do? Well... Our problem here is only FREEZE of pppoe servers on wireless.

We have actually ONE device running ROS 3.13 freezing due to PPPOE server problem, so... If you want, we can give you full access to our router and also, we can set any LOGs you may need to a remote Syslog right inside MTs lab. Also, we recommend that MT keep monitoring this device with SNMP as it is the way we detect this freeze thing over the time. Otherwise you must learn yoga and sit watching the box til the freeze happens. Is there anything else we can do to aid MT team?

Thank you
Ozelo

Problem with PPPoE is in RouterOS version which is not RC (release candidate).
If you (I) use 3.0rc9 for example, on RB333/600, pppoe never freeze and everything works fine. Ok, try to upgrade it to 3.13 - pppoe die.
I tried to do same thing (put 3.0rc9) on RB433 but not working on this board, only 3.0 and up.
BUT....RB333 is end of life and now what ?!

Problem with PPPoE is in RouterOS version which is not RC (release candidate).
If you (I) use 3.0rc9 for example, on RB333/600, pppoe never freeze and everything works fine. Ok, try to upgrade it to 3.13 - pppoe die.
I tried to do same thing (put 3.0rc9) on RB433 but not working on this board, only 3.0 and up.
BUT....RB333 is end of life and now what ?!

Yeah, I think the same.Tomorrow I will go to replace the failing RB333 with an RB433 ROS 3.13, I think that the problem will disappear.
Another consideration for Normis: you have all our supout files so would be easy for you to check our configurations ( pppoe setup, mrru etc ), so why you said:
"compare your setups - do you have big PPPoE server with lots of users or small router over wireless - those are completely two different things. Do you use RADIUS, queues, MRRU??
Have you enabled pppoe,debug logging to remote place to actually see what was the last thing your pppoe concentrator did?"
You should know our configurations and compare them, is it thrue?
With best regards to all
Alessandro

Boys
As I could observe, the problem is given in some clients sporadically, nevertheless the pppoe server continues working well in the rest of the clients. The pppoe server always works, except that in some clients it(he,she) freezes.

Regards

You should know our configurations and compare them, is it thrue?

that is my point - all your cases are completely different. "pppoe disconnects" is not one specific problem. it can be caused by anything

You should know our configurations and compare them, is it thrue?

that is my point - all your cases are completely different. "pppoe disconnects" is not one specific problem. it can be caused by anything

Ok Normis, I agree with you, I know what does mean " problem determination", but seems that after a lot of customers claiming a problem on PPPoE, no actions or a precise direction was taken. PPPoE problem are arised since ros 2.9 and my impression is that no actions to correct the problem or at least "some" problems were made, am I wrong? Today I will go to change RB333 with RB433, I will let you informed on results, obviously I will not change the clients; another info for you: yesterday I downgraded to version 3.0 rc14, the PPPoE login of clients was EXTREMELY fast compared to version 3.13 ( 5 seconds - 30 seconds ) and disconnections are decreased a lot, another strange think is that disconnections starts at around 10:00 local time ( 08:00 gmt ), it maybe that the temperature of RB333 increases and the problem arises? It maybe a solution to decrese the clock to 266Mhz?
Thanks and regards
Alessandro

that is my point - all your cases are completely different. "pppoe disconnects" is not one specific problem. it can be caused by anything

sorry but i think you did not read our posts . the problem is not "pppoe disconnects". the problem is : user is connect but no transfer and when user disconnect by hand and again connect the problem solves until next time . next time may be 30 seconds later or 1 day later.

The problem is solved when layer2 requests , already it is by means of neighbor, mac telnet, mac-winbox, etc. Of so many things that probe, a temporary, serious solution to arm(assemble) a programmed script who does a mac-ping every 5 second to the mac of every client, forcing to routeros to continue the communication in the layer2. The topic is that from console I believe that the mac-ping is not supported. We hope that this problem can be fixed, since our clients complain and it worsens the image of our company.

saludossss

I am not currently seeing this problem but I will throw a few observations in here. If I have a wireless client with a real crudy connection that drops a lot of packets I will see this problem where the PPPoE connection freezes. Typically you need to drop and restart the PPPoE client to solve it. Never looked into it extensively because a client with a very poor wireless connection like that needs it fixed anyway.

Perhaps thats what is being seen here?

Matt

It is supposed that on having done downgrade I lose the whole configuration? Maybe the momentary solution this in to use her 3.0rc14 ... but theoretically we are losing all the propernesses up to(even) her(it) v3.13

Regards

pd: 3.0rc14 runs well on rb433ah? I have understood that alone works from the version 3.0

Hello
I'm having the same PROBLEM!!! and I said its in other post!!. Is exactly like all of you say, I client can work perfect and when it start have a lot of connections, the PPPoE Server close that connections, and the only way to solve this problems is to reconnect the user, but imagine, to explain it to all you client, they will kill as!!
Today I take our my PPPoE Server, that was an IBM Server, with all you want. And I remplace its with a RB 433 AH, and I solved that problems. The pppoe server doesn´t die, but a new problem appears. Its start assing IP that are used in other clients!! I solve a problem to go to another!!. The post of that problem is:

http://forum.mikrotik.com/viewtopic.php?f=2&t=26486

For that that have the problem of PPPoE Die, please try changing the server for some RB. In my experience I have a RB433 AH, without change, with 130 simoultaneous clients, and the processor is in 35%, but with IP problems!!!!
Well I hope it can help to solve the problem to MKT team, and MKT users
Thank, and if somenone know hao to solve the IP problems, please help me
Rafael Lore

I believe that it is not a problem of hardware. I have one rb433ah and one rb532r5 both with the version 3.13 and the two has the same problem. The pppoe client dies. It covers some perfectly, and others must reconnect. We say that to work with ip fixes in the troubled clients up to(even) so much one finds a solution. This produces a security breach, since we have to assign ip hinge to interface wireless and generally not all the clients support wpa. For which it is necessary to work without encript or with wep that already we know poor that is.

saludosss

Hello, I made some changes but the problem is still present but I've noticed some special things:
around 4 to 5 clients remain connected all the time ( 24Hours now) other disconnects frequently, but the clients that today remained connected yesterday did NOT, it seems that the first 4 to 6 clients that will connect first, remain connected without problems, independently of client brand or setups.
Now I am going to disconnect all clients and reconnect them manually in another order; the disconnections takes place in the hottest ours of the day ( from 11 to 16 local time )
Regards
Alessandro

No no!!
This reemplace only extended the agony by three hours, and the crashed!!! If I restarted the RB, died again, so I came back to my old configuration!!
The problems appears to me, when I use Public IP Address, but if I give my client Private IP Address and then make a masquerade in the Server, it works PERFECTLY!! But whrn I tried to work with Public IP Address it crash!!! I clarified that I'm ussing in both case the RouterOS v 3.13
This problems appear to you when you are working with public or private IP?
He Normis, its is possible that can be a problems in this version of RouterOS, or we contineous trying to solve it?
Thanks
Rafael Lore

Hello
I'm having the same PROBLEM!!! and I said its in other post!!. Is exactly like all of you say, I client can work perfect and when it start have a lot of connections, the PPPoE Server close that connections, and the only way to solve this problems is to reconnect the user, but imagine, to explain it to all you client, they will kill as!!
Today I take our my PPPoE Server, that was an IBM Server, with all you want. And I remplace its with a RB 433 AH, and I solved that problems. The pppoe server doesn´t die, but a new problem appears. Its start assing IP that are used in other clients!! I solve a problem to go to another!!. The post of that problem is:

http://forum.mikrotik.com/viewtopic.php?f=2&t=26486

For that that have the problem of PPPoE Die, please try changing the server for some RB. In my experience I have a RB433 AH, without change, with 130 simoultaneous clients, and the processor is in 35%, but with IP problems!!!!
Well I hope it can help to solve the problem to MKT team, and MKT users
Thank, and if somenone know hao to solve the IP problems, please help me
Rafael Lore

As i said before , this problem occurs very randomize . maybe it does not happen in a day and some times it happen each 30 seconds . so if you think you could solve the problem , please wait 2 or 3 days and if problem really does not came back then say your solution here .

3.14rc1 is out now

3.14rc1 is out now

Thanks but nothing about pppoe problems in Changelog . so what does it mean ????!!!!

Normis ¿ they have patched the kernel? If it is not this way, that are extracted the new version, it was sounding nicely but it does not change anything. That seems to be troublesome, but it is worrying to have this problem and mas when there are people who pays for our services.

Regards

That is true!! there is not solution to ower problem!! If you want normis I give full access to a MKT that present its problems. I think that the problems is the communication between MKT and Radius Server
Thanks
Rafael Lore

Already probe disabling the autentificacion radius and the accounting from ppp aaa and nevertheless the problem persists, I believe definitively that the current kernel has problems with the tunnels ppp.

regards

pd: rafa mi msn es lucho_012@hotmail.com ... saludos
Santa Cruz
Caleta Olivia

My setup

Ros: v3.11
Intel Pentium Dual Core: 3000 Mhz (set multi-cpu =YES)
RAM= 1Gb
PPPoE Secrets = 372.
Licencie = Level 5
Max simultaneously PPPOE users = about 211

I've not major problems. Everything is OK, but sometimes
%CPU overhead (each 2 o 3 days), and PPPoE server begins
to disconnect users or get random fails (but no Ros CRASH).

I restart ROS when this happens, and everything is OK for
2 or 3 days.

I'm not using RADIUS over my PPPoE server, but I want to
use FREE RADIUS soon!

best regards
andres.

My setup

Ros: v3.11
Intel Pentium Dual Core: 3000 Mhz (set multi-cpu =YES)
RAM= 1Gb
PPPoE Secrets = 372.
Licencie = Level 5
Max simultaneously PPPOE users = about 211

I've not major problems. Everything is OK, but sometimes
%CPU overhead (each 2 o 3 days), and PPPoE server begins
to disconnect users or get random fails (but no Ros CRASH).

I restart ROS when this happens, and everything is OK for
2 or 3 days.

I'm not using RADIUS over my PPPoE server, but I want to
use FREE RADIUS soon!

best regards
andres.

also see this another topic before doing the job unregister_netdevice problem . maybe it stops you .

hello I installed ROS 3.14rc1 but the problem is still present, but I noticed the disconnections happens MORE frequently than before so I immediately downgraded to ROS 3.13.
For your info I changed previous RB333 with RB433 with no success, same problem.
Regards
Alessandro

Hello Mtik team, I am thinking about that maybe the Virtual AP on the same interface of PPPoE that is causing the problem.
I've noticed the overall ACK timeout of the interface decreasing from about 350 usec to 50 usec if I disable the VAP interface; at the same time the overall tx CCQ increases from about 30/60% to about 90%.
At this point I left VAP disabled and now I am watching if some PPPoE client disconnects, after about 2 ours nobody has disconnected, I will let you informed.
Regards
Alessandro

Maybe in your case the virtual ap is an attenuant. At my case I am not being employed with th and nevertheless I have the problem that they all.

regards

Hello
I don't have any VAP, in the server where I have PPPoE, is only PPPoE Server with a public interface and an internal interface. I have done this configuration to other people and works perfectly!!!. And when I do to my network its CHASH . I think that are some situation that occurs that cause its. So it will be extremelly difficult to found!!!
The different that happened to me when I try v 3.11 and 3.13, is that with 3.13 didn't crash, but it have problem to assing IP, the server tried to assing IP that was in use in other user, the server didn't crash, but some clients couldn't connect. So I prefered to be near the server to restarted it, that have some clients called me every minute!!
And something that I don't understand, is with this happen me when I use public IP Address. If I use private IP Address, with NAT, works PERFECTLY!!, but I want to use my range of Public IP!!! I'm paying for its IP!!!!
And Andres do you remember that I have that problem that you have when I configured my Server with multi-cpu =YES, so I disabled its, and solved my problems. And in different of the CPU used with its and no, was nothing!! So I recommend to disable its!!
Thanks
Rafael Lore

Hi guys, I found this topic http://forum.mikrotik.com/viewtopic.php?f=1&t=19491;
it seems related to my problem, with vap enabled overall tx ccq drops below 40%, all clients connected to normal interface may disconnect;
disabling VAP, tx ccq rises above 90% ( wireless card R52H - mode G ).
only 1 user connected to VAP/Hotspot and 16 users connected to WLAN in PPPoE mode.
Regards to all
Alessandro

Hello, problem is not solved yet, does anyone or Mtik team has some news?
Regards
Alessandro

without news for now

Sorry cuz it is a bit OT, but today Ive found for the first time something different. A customer complaining about throughput. Just disabled and enabled one single mangle rule related to connection limit and then the user was not passing through there. In other words, mangle rules can stuck and was forcing all that customers through it, completely ignoring the rule matches. I wonder how many scripts I need to create to periodically refresh (disabling and re-enabling) those hundred mangle rules due to this randomness issues. It was a ROS 3.4 and yes, pppoe still freeze on 3.14rc1 without any good news from MT about that.

Got this OSPF error message on the logging facility atm of the freeze stuff:

17:37:14 route,ospf,error Cannot send packet: setsockopt IP_MULTICAST_IF failed
17:37:14 route,ospf,error ifindex=15
17:37:14 route,ospf,error error=Cannot assign requested address

And then the router is dead for about 30 minutes til it get ride of its frosty state. Yes, ROS 3.14rc1

Got this OSPF error message on the logging facility atm of the freeze stuff:

17:37:14 route,ospf,error Cannot send packet: setsockopt IP_MULTICAST_IF failed
17:37:14 route,ospf,error ifindex=15
17:37:14 route,ospf,error error=Cannot assign requested address

And then the router is dead for about 30 minutes til it get ride of its frosty state. Yes, ROS 3.14rc1

the freeze topic is this one http://forum.mikrotik.com/viewtopic.php?f=2&t=25567 . please move or copy your post there . it is better for finding the solution .

i did n't face this type of problem before, But when i upgrade v2.9.51 to v3.13 i m really confusing abt "Peer is not responding"


is it cable problem or router????

I have a simal problem. When configuring a RB411, wlan as station pseudobridge and then brigde wlan and eth, so that I then create a PPPoE tunnel from a PC, the connection is stable and works fine. However if I use a Senoa SL-2611CB3 Plus and connect to the network in the same manner (PPPoE), the connection suddenly dies (the PPPoE is still active, but no traffic passes). The time interval before it freezes/dies is not fixed. This morning I established a connection, left the machine and after 1 hour returned to find the connection still working. However after starting to transfer data the connection froze after a while. So if no traffic is transferred the PPPoE connection stays active, but as soon as data is transferred it freezes after a while (only with Senao).

The access point is a RB-433AH and a RB52H, signal strength of the RB-411 and Senao -85dBm.

any news? i downgrade 3.13 , 3.14, 3.15 (works on rb532) to 2.9.51 and pppoe server works well.

We encounter the same issue. Here is a background on our network:

Over 550 MT clients on 3.2-3.10 combination of RB133 and 411's. using R52h's and SR5.
All WDS Bridged network to MT core router on a DUAL CORE Dell Server
Microsoft Enterprise Server runnig IAS for Radius.
Local PPPOE client running on MT client boards

We have several clients that have the disconnect symptom... this week I decided to spend considerable time on one.
I replaced their unit 3 times... this week. I replaced their power supply and injector 3 times and I put a temporary cable in a completely different route.... It made no difference. Signal strength is -66/-66 with CCQ close to 98%. We NEVER see a wireless disconnect and performance is flawless. We do run NSTREM on our PTMP AP's

On the AP I deleted their WDS registration and added again. We use VLANS for Management and WWW traffic, we replaced the ip's on the client.

No change..... same problem ... very frustrating!

My conclusion... some sort of untraceable wireless interferance that effects the connection or some sort of weird AP related configuration glitch. Any other helpful suggestions are appreciated....

Anyone know how many clients you can serve on a single radius server... is there a limit???

Please resume this topic here : http://forum.mikrotik.com/viewtopic.php?f=2&t=28422 . thanks