Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 11 Jan 2012 03:05:29 +0400
From:      Lev Serebryakov <lev@FreeBSD.org>
To:        FreeBSD current <freebsd-current@freebsd.org>
Subject:   Very fresh (two days ago) 10-current becomes completely unresponsive under load
Message-ID:  <1791250845.20120111030529@serebryakov.spb.ru>

next in thread | raw e-mail | index | archive | help
Hello, FreeBSD.

  I have home router+AP based on Soekris net5501 (500Mhz GEODE CPU,
 512MiB of RAM).

  It is equipped with vrX NICs (4 of them on-board, one is used as
 upstream to my ISP, other are unconnectd), em0 PCI board (legacy
 driver, downstream to my home network) and PCI ath0 (WiFi, for my
 home wireless network).

  ISP is connected with PPPoE, mpd 5.6 is used. It is shaped to 40Mbit
 by ISP. Alos, I have IPv6 tunnel to Hurricane Electric.

    Box does PPPoE/ipfw/ipfw-nat, DNS (bind from base system), dhcpd,
 hostapd, nut (network client for UPS, UPS is connected to other box)
 and it's all.

  It has serial console to my other computer (no VGA//KBD console on
 this hardware).

  I'm running 10-CURRENT to have access to latest ath (WiFi) code.

  I've used version from 23 Nov 2011 till yesterday, without any
 significant problems.

  Then it was upgraded to sources csupped 9 of Jan 2011.

  And now under network load (it could be loaded up to 75% CPU
 according to my previous experience, I never seen more than that), it
 becomes totally unresponsive. It could be pinged. But SSH connection
 takes tens of minutes. Typing login and wassword on serical console
 could take 2-5 minutes too. DNS timeouts become frequent. It is
 alive. But unresponsive. I even could not run "top" on it in such
 moments, as it doesn't response to typing, even if I logged in before
 this problem. After some time it answers and works normal for 2-3
 minutes (ssh, serial console, etc.) but after that it "hangs" again.

  No any special messages on console or in demsg output.

  OH! I have top running right now, when it "hangs". 0% idle time, LA
 becomes 20 when it have only 35 processes at all, but there is no specific
 process consuming CPU.

  Kernel config is stripped down GENERIC, with SHED_ULE.

  And I never have such problems with 10-CURRENT from 23 Nov 2011.

--=20
// Black Lion AKA Lev Serebryakov <lev@FreeBSD.org>




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1791250845.20120111030529>