Bug#572835: enhancement: start bridge using ifup faster
On Sun, 07 Mar 2010 12:04:36 +0600 Mikhail Gusarov dotted...@dottedmag.net wrote: Twas brillig at 17:17:51 06.03.2010 UTC-08 when shemmin...@vyatta.com did gyre and gimble: ifup snippet for bridge-utils uses 'sleep 1' in loop in order to wait for bridge. This timeout makes 'ifup br0' really slow, 1.2s in my setup. Attached patch makes ifup snippet use 'sleep 0.01' if fractional secons are implemented in sleep, thus making ifup much faster (0.16s as measured). SH I don't think this sleep is needed at all. I tested it - a bit of delay is still necessary. 0.01 is always enough in my configuration though, but there may be slower machines. Maybe there is a way to wait for bridge instead of polling it existance, but I'm not aware of it. I think this is related to forwarding delay. Ideally forwarding delay should be off unless STP is enabled. Looks like Ubuntu script reads that value of forwarding delay and adjusts based on that. -- -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#572835: enhancement: start bridge using ifup faster
Package: bridge-utils Version: 1.4-5 Severity: wishlist ifup snippet for bridge-utils uses 'sleep 1' in loop in order to wait for bridge. This timeout makes 'ifup br0' really slow, 1.2s in my setup. Attached patch makes ifup snippet use 'sleep 0.01' if fractional secons are implemented in sleep, thus making ifup much faster (0.16s as measured). P.S: never mind the Ubuntu info, that's just the desktop I'm reporting bug from. -- System Information: Debian Release: squeeze/sid APT prefers karmic-updates APT policy: (500, 'karmic-updates'), (500, 'karmic-security'), (500, 'karmic') Architecture: i386 (i686) Kernel: Linux 2.6.32-9-generic-pae (SMP w/2 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Versions of packages bridge-utils depends on: ii libc6 2.10.1-0ubuntu16 GNU C Library: Shared libraries Versions of packages bridge-utils recommends: ii ifupdown 0.6.8ubuntu21 high level tools to configure netw bridge-utils suggests no packages. -- no debconf information --- /var/lib/lxc/test/rootfs/usr/share/bridge-utils/ifupdown.sh 2008-08-28 04:52:20.0 +0700 +++ ifupdown.sh 2010-03-07 03:45:03.650307276 +0600 @@ -213,13 +213,16 @@ then /bin/echo -e \nWaiting for $IFACE to get ready (MAXWAIT is $MAXWAIT seconds). +# Use 0.01 delay if available +sleep 0.01 2/dev/null COUNT=$(($COUNT * 100)) + unset BREADY unset TRANSITIONED COUNT=0 while [ ! $BREADY -a $COUNT -lt $MAXWAIT ] do - sleep 1 + sleep 0.01 2/dev/null || sleep 1 COUNT=$(($COUNT+1)) BREADY=true for i in $(brctl showstp $IFACE|sed -n 's/^.*port id.*state[ \t]*\(.*\)$/\1/p')
Bug#572835: enhancement: start bridge using ifup faster
On Sun, 07 Mar 2010 03:50:18 +0600 Mikhail Gusarov dotted...@dottedmag.net wrote: Package: bridge-utils Version: 1.4-5 Severity: wishlist ifup snippet for bridge-utils uses 'sleep 1' in loop in order to wait for bridge. This timeout makes 'ifup br0' really slow, 1.2s in my setup. Attached patch makes ifup snippet use 'sleep 0.01' if fractional secons are implemented in sleep, thus making ifup much faster (0.16s as measured). P.S: never mind the Ubuntu info, that's just the desktop I'm reporting bug from. -- System Information: Debian Release: squeeze/sid APT prefers karmic-updates APT policy: (500, 'karmic-updates'), (500, 'karmic-security'), (500, 'karmic') Architecture: i386 (i686) Kernel: Linux 2.6.32-9-generic-pae (SMP w/2 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Versions of packages bridge-utils depends on: ii libc6 2.10.1-0ubuntu16 GNU C Library: Shared libraries Versions of packages bridge-utils recommends: ii ifupdown 0.6.8ubuntu21 high level tools to configure netw bridge-utils suggests no packages. -- no debconf information I don't think this sleep is needed at all. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#572835: enhancement: start bridge using ifup faster
Twas brillig at 17:17:51 06.03.2010 UTC-08 when shemmin...@vyatta.com did gyre and gimble: ifup snippet for bridge-utils uses 'sleep 1' in loop in order to wait for bridge. This timeout makes 'ifup br0' really slow, 1.2s in my setup. Attached patch makes ifup snippet use 'sleep 0.01' if fractional secons are implemented in sleep, thus making ifup much faster (0.16s as measured). SH I don't think this sleep is needed at all. I tested it - a bit of delay is still necessary. 0.01 is always enough in my configuration though, but there may be slower machines. Maybe there is a way to wait for bridge instead of polling it existance, but I'm not aware of it. -- http://fossarchy.blogspot.com/ pgp3NaerAKgHV.pgp Description: PGP signature