So I had put accel-ppp live with heavy load

PPPoE related questions
dimka88
Posts: 866
Joined: 13 Oct 2014, 05:51
Contact:

Re: So I had put accel-ppp live with heavy load

Post by dimka88 »

No, I think for 32. Can you show `perf top` screenshot. What in accel-ppp logs and dmesg? Provide please also `accel-cmd show stat`
hashbang
Posts: 135
Joined: 12 Jul 2015, 10:28

Re: So I had put accel-ppp live with heavy load

Post by hashbang »

dimka88 wrote: 02 Mar 2020, 19:36 No, I think for 32. Can you show `perf top` screenshot. What in accel-ppp logs and dmesg? Provide please also `accel-cmd show stat`
ty,
will post when problem comes again.
hashbang
Posts: 135
Joined: 12 Jul 2015, 10:28

Re: So I had put accel-ppp live with heavy load

Post by hashbang »

i think the problem is this. frr user using ping command lots of time. Dont know why frr (zebra and ospfd) using ping so much.
Screenshot from 2020-03-05 12-58-13.png
Screenshot from 2020-03-05 12-58-13.png (184.96 KiB) Viewed 5282 times
thnx
dimka88
Posts: 866
Joined: 13 Oct 2014, 05:51
Contact:

Re: So I had put accel-ppp live with heavy load

Post by dimka88 »

Hi @hashbang, I don't know about frr ping command, but I think you can make some things. Try set in section [ppp] param unit-cache=3000 and then execute `accel-cmd reload`
from docs https://accel-ppp.readthedocs.io/en/lat ... n/ppp.html
unit-cache=n
By default is disabled: unit-cache=0

Specifies number of interfaces to keep in cache. It means that don’t destroy interface after corresponding session is destroyed, instead place it to cache and use it later for new sessions repeatedly. This should reduce kernel-level interface creation/deletion rate lack.
hashbang
Posts: 135
Joined: 12 Jul 2015, 10:28

Re: So I had put accel-ppp live with heavy load

Post by hashbang »

Hi,
here is a screenshot I forgot earlier to post. It show high cpu usage by accel-pppd + ping to ospf network by user frr. The first row showing accel-ppp consuming 78% cpu No idea is it accel-ppp or frr problem.
Screenshot from 2020-03-05 15-33-59.png
Screenshot from 2020-03-05 15-33-59.png (234.02 KiB) Viewed 5173 times
dimka88
Posts: 866
Joined: 13 Oct 2014, 05:51
Contact:

Re: So I had put accel-ppp live with heavy load

Post by dimka88 »

Did you check journalctl and dmesg? What about accel-ppp.log and accel-cmd show stat?
hashbang
Posts: 135
Joined: 12 Jul 2015, 10:28

Re: So I had put accel-ppp live with heavy load

Post by hashbang »

i think this problem is caused by frr. Next time i'lll check all the parameters and post
ty
dimka88
Posts: 866
Joined: 13 Oct 2014, 05:51
Contact:

Re: So I had put accel-ppp live with heavy load

Post by dimka88 »

As alternate routing daemon, you can use `bird`, that is an idea to confirm/decline that issue exactly with FRR
hashbang
Posts: 135
Joined: 12 Jul 2015, 10:28

Re: So I had put accel-ppp live with heavy load

Post by hashbang »

Hi
ty,
Somemore inputs again had problem today
journalctl doesnt show anything worth to be mentioned
dmesg all showing pppoe renamed to xxx
but something more to be given importance 30k connections on finish states
accel-cmd show sessions | grep finish | wc -l
30363

I just restarted the services

screenshots attached :
Screenshot from 2020-04-28 12-29-10.png
Screenshot from 2020-04-28 12-29-10.png (252.28 KiB) Viewed 5162 times
Screenshot from 2020-04-28 12-29-25.png
Screenshot from 2020-04-28 12-29-25.png (110.63 KiB) Viewed 5162 times
dimka88
Posts: 866
Joined: 13 Oct 2014, 05:51
Contact:

Re: So I had put accel-ppp live with heavy load

Post by dimka88 »

Try to set '[radius]acct-timeot=0` and server=.....req-limit=0 in case if rodius does not responding.
Can you also provide your [radius] section?
Post Reply