Junos
Highlighted
Junos

MX104 PPPoE Issue

a week ago

We have an MX104 that had pfed hit max memory usage and that in turn caused service interruption to PPPoE users trying to authenticate to the BNG.  I tried restarting the statistics-service (pfed) and that freed up the memory, but that did not resolve the issue.  I've tried to clear out all the subscribers as all their information in the database (show subscribers) is still there.  None of that info is clearing out and the pp0.xxx interfaces are still showing up.  Also, in the log I am getting:

 

Dec  6 10:04:29  glttwy-fnb-mx104-1 authd[46609]: ../../../../../../src/junos/usr.sbin/authd/plugin/radius/authd_plugin_radius_module.cc:1566 Failed to get SDB snapshot for session-id:1220550
Dec  6 10:04:29  glttwy-fnb-mx104-1 authd[46609]: ../../../../../../src/junos/usr.sbin/authd/plugin/radius/authd_plugin_radius_module.cc:1566 Failed to get SDB snapshot for session-id:1227095
Dec  6 10:04:29  glttwy-fnb-mx104-1 authd[46609]: ../../../../../../src/junos/usr.sbin/authd/plugin/radius/authd_plugin_radius_module.cc:1566 Failed to get SDB snapshot for session-id:719925
Dec  6 10:04:29  glttwy-fnb-mx104-1 authd[46609]: ../../../../../../src/junos/usr.sbin/authd/plugin/radius/authd_plugin_radius_module.cc:1566 Failed to get SDB snapshot for session-id:1222408

 

And also:

 

Dec  6 10:16:29  glttwy-fnb-mx104-1 cosd[1845]: COSD_GENCFG_WRITE_FAILED: GENCFG write failed (op, minor_type) = (add, classifier to ifl) for tbl 9 if 641 pp0.1074000664 Reason: File exists
Dec  6 10:16:29  glttwy-fnb-mx104-1 cosd[1845]: COSD_GENCFG_WRITE_FAILED: GENCFG write failed (op, minor_type) = (add, classifier to ifl) for tbl 13 if 641 pp0.1074000664 Reason: File exists
Dec  6 10:16:29  glttwy-fnb-mx104-1 cosd[1845]: COSD_GENCFG_WRITE_FAILED: GENCFG write failed (op, minor_type) = (add, ifl tcp) for tbl 2111118723 if 641 pp0.1074000664 Reason: File exists

 

 

Is there anyway we can get this box revived without having to reboot?  Version is 14.2R8.4.

13 REPLIES
Junos

Re: MX104 PPPoE Issue

a week ago

Hi,

 

As you already aware, 14.2 is not subscriber-management release. You will end up with DB related issues frequently.

Try restarting authd and jpppd. 

 

restart general-authentication-service

restart ppp-service

 

Regards,

Rahul

Junos

Re: MX104 PPPoE Issue

a week ago

I have bought an MX204 that will take over on these and that's literally going to happen like next week, but of course this box had to crash again just before....

 

I restarted both of those, but I am still seeing all these subscribers listed in "show subscribers" when they certainly are not supposed to be there.

Junos

Re: MX104 PPPoE Issue

a week ago

How many subscribers are there? can you please share the show subscribers summary output.

Junos

Re: MX104 PPPoE Issue

a week ago

I've moved all the subscribers (they are terminated via PWHT) to a secondary BNG.  So technically there should be none on this box currently.

 

Subscribers by State
   Configured: 11
   Active: 110
   Terminating: 634
   Total: 755

Subscribers by Client Type
   DHCP: 91
   VLAN: 31
   PPPoE: 633
   Total: 755

Junos

Re: MX104 PPPoE Issue

a week ago

Ok.  Most of them stuck in terminating.

 

try restarting subscriber-management

Junos

Re: MX104 PPPoE Issue

a week ago

Restarted.  Didn't seem to do anything:

 

Subscribers by State
   Configured: 11
   Active: 110
   Terminating: 634
   Total: 755

Subscribers by Client Type
   DHCP: 91
   VLAN: 31
   PPPoE: 633
   Total: 755

Junos

Re: MX104 PPPoE Issue

a week ago

Try restarting auto-configuration. In case this doesn't help then reboot is the only option.

Junos

Re: MX104 PPPoE Issue

a week ago

Reboot it is then...

Junos

Re: MX104 PPPoE Issue

a week ago

Perhaps I'll try upgrading this to 17.4 as well -- I know that in order to do that, I have to enable a subscriber-mangement subsystem and set the max-db-size.  What would be the recommended setting there?  Are there any other caveates with sub management?

 

In order to go to 17 from 14, should I first go to 15 then to 17 or can I jump right to 17?  What specific 17 version would be recommended on the MX104?

Junos

Re: MX104 PPPoE Issue

a week ago

MX104, max-db-size is 100m

 

set system configuration-database max-db-size 100M

set subscriber-management enable

 

Please upgrade to 15.1 and enable above two knob. Reboot the complete chassis i.e. both routing engine.

Then upgrade the box to 17.4 and reboot chassis.

 

From 15.1, both the RE will be HA mode.

 

https://www.juniper.net/documentation/en_US/junos/topics/task/configuration/subscriber-management-en...

 

Regards,
Rahul Nayar

Junos

Re: MX104 PPPoE Issue

a week ago

Recommended version is 17.3R3-S2

 

Download Results for:   MX104

Select:
 
Junos

Re: MX104 PPPoE Issue

a week ago

Are there any other gotchas you can think of going from 14.2 to 17.3?   In addition to PWHT+PPPoE, we do basic IP terminiation, basic l2circuit psuedowires and a few VPLS instances.  Is there anything you can think of going from 14 to 17 that has changed that I will need to update?

Junos

Re: MX104 PPPoE Issue

a week ago

There is archietecture change in NG subscriber-management release. PWHT is popular feature and there is no caveat in 17.3