mikrotik - pppoe client delay connection [SOLVED]

RouterOS General
1

I have one Mikrotik Routerboard with 6.41.3 firmware. VLAN for multiple areas are configured in Cisco 3560 Gigabit switches and TRUNK is connected with CCR. PPPoE Server with proper settings is configured for each vlan.

RB resources are OK, no heating , CPU usage 3 to 4 % max , PING from users PC (and most have wifi routers) to Server is Okay with 0 delay.

Problem is that at random time when user tries to dial pppoe, he keep stucked on DIALING (in the pppoe dialer) … then after long pause its goes to verifying username and password and then connected after another long pause or some times fails. It happens randomly for multiple users wither wired desktop user or user with wifi router.

How can I trace what is wrong?

2

Sniff traffic with wireshark or enable PPPoE debug on client side, I assume there is some broadcast issue. PADI is sent to L2 broadcast address, and maybe it gets dropped somewhere on your router on stick configuration. Broadcast storm control ?

3

Mikrotik CCR 1036 - V 6.41.3
There is no Filter/NAT/Mangle rule. All users gets public ip and route to upstream ISP.
Problem is most times when users dial/redial they gets long pause on dialing seems they are unable to find the PPPoE Server.

4

I got it form your first post. Like i said, it could be some broadcast limit. And for reconnection, maybe you have old sessions still up?

5

I dont see any error in mikrotik. infact I feel that mikrotik is not receiving the PADI requests , since its live network with around 2k users so i cannot enable ppp debug.

For testing I created local ID in secrets to avoid any radius related issue. but still its same.
Then for further testing I connected my test PC and laptop directly to mikrotik empty port & tried to connected , & it behaved same. sometimes it connects instantly but most times it takes times to connect.
is there any setting I can check in mikrotik ?

6

Beside MTU/MSS there is not much to configure. Its unclear why would you get slow dial time while directly connected to router. Only option is “PADO delay” inside PPPoE server if you have set by accident.
Connect PC again and torch that interface. Then post what you got.

7

Do you use compression in ppp profile? If yes, disable it. What type of authorization you use in PPPoE server? Try only chap.

8

@blajah
this RB is in production with 2000 active users, therefore running torch and look for specific user is quite hard.

@Anumrak
PAP/CHAP are selected.
Compression is on DEFAULT. I will disable it.

I will change compression setting and will let u know.

9

I tried every setting possible but still having problem that some times users connect instantly, but most of times it gets stucks on dialing for long time and then connect.

Ok yesterday I tried following. create pppoe server on another port on CCR, connect test client pc directly to this port. Ran the TORCH on this interface and then tried to dial the pppoe dialup connect on test pc.

When client was stuck on DIALING, & In TORCH, I was seeing PPPOE Discovery request but it was not going further,
so why its stucking to PPPoE Discovery most of the time?

10

Solved.
When I lower the RADIUS TIMEOUT value from 3000 ms to 500-700 ms and the problem got solved. The user is now able to connect instantly or within 5 seconds.

I used this value at almost all other systems but this particular one, radius timeout value works only if under 700 ms no matter ier is local or radius. I thought mikrotik always looks in SECRETS first then radius if configured, but here different experience have been faced.