Development discussion of WireGuard
 help / color / mirror / Atom feed
From: Luiz Angelo Daros de Luca <luizluca@gmail.com>
To: wireguard@lists.zx2c4.com
Subject: Sys CPU usage increasing over time
Date: Tue, 16 Jan 2018 23:24:42 +0000	[thread overview]
Message-ID: <CAJq09z7WRCpOgF24QztfY2++C4tVD2MkN_w3YLoQpmDtiuh5Dw@mail.gmail.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 1979 bytes --]

Hello,

I'm testing wireguard with two LEDE (OpenWrt) 17.01.4 nodes. The "server"
is a TL-WR2543ND with 64MB RAM and the "client" (the one with a defined
endpoint address) a poor TL-WR741ND v4 with 32 RAM.
Both are ar71xx systems, arch MIPS 24Kc CPU, with little to no performance
difference.

As TL-WR741ND has only 4 MB or flash, the smaller footprint of wireguard
allowed me to squeeze all needed packages without removing any original
package (as web gui) creating a custom image. That was unimaginable with
openvpn. Thanks wireguard!

Both are using a little bit outdated wireguard module because modules are
only rebuild on new dot releases. Userland can be updated easier.

kmod-wireguard - 4.4.92+0.0.20171017-1
wireguard - 0.0.20171221-1
wireguard-tools - 0.0.20171221-1

Everything seems to work as expected. However, on the client (TL-WR741ND),
I noticed that something between half a day to a couple of days normal
network usage (not VPN) gets slower and slower. Checking the device,
loadavg reaches values as 15 a little bit before the router stops to work
and I need to do a cold restart. top tells me sys usage eats 90% of CPU. If
I restart wireguard interface, nothing changes. The CPU falls drastically
if I remove wireguard module from kernel (rmmod), returning to a normal
loadavg of 0.05.

My VPN link is normally idle, using a couple of MiB per day. The VPN usage
seems to not affect the time needed to manifest the problem, but normal
network usage does seem to make the problem appear sooner.

The strange thing is that the "server" node does not show the same
behavior. Maybe it is related to the fact that it has currently no local
network usage there. Both nodes uses the same wireguard binary code.

Is this a known issue of version 0.0.20171017? Or do I get something new? I
tried to read all changelog since 20171017 but I could not identify any
obvious fix.

Is there anything I can do?

Regards,


-- 

Luiz Angelo Daros de Luca
luizluca@gmail.com

[-- Attachment #2: Type: text/html, Size: 2512 bytes --]

             reply	other threads:[~2018-01-16 23:21 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-01-16 23:24 Luiz Angelo Daros de Luca [this message]
2018-04-16 17:13 ` Jason A. Donenfeld
2018-04-16 18:35   ` Luiz Angelo Daros de Luca

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CAJq09z7WRCpOgF24QztfY2++C4tVD2MkN_w3YLoQpmDtiuh5Dw@mail.gmail.com \
    --to=luizluca@gmail.com \
    --cc=wireguard@lists.zx2c4.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).