HZ in RELENG_5? tcp_subr.c related
Dear all, I read this commit message to RELENG_5: http://www.freebsd.org/cgi/getmsg.cgi?fetch=436074+438355+/usr/local/www/db/text/2005/cvs-all/20050206.cvs-all and commented out the HZ=1000 line in my kernel config so I use the default. But I see only 100 interrupts/sec on clk (with systat) so is it true that the default HZ has changed from 100 to 1000 in RELENG_5? Thanks, -Harry pgpRDXkmcPZYT.pgp Description: PGP signature
Re: HZ in RELENG_5? tcp_subr.c related
On Tue, 22 Feb 2005, Emanuel Strobl wrote: > Dear all, > > I read this commit message to RELENG_5: > http://www.freebsd.org/cgi/getmsg.cgi?fetch=436074+438355+/usr/local/www/db/text/2005/cvs-all/20050206.cvs-all > and commented out the HZ=1000 line in my kernel config so I use the default. > But I see only 100 interrupts/sec on clk (with systat) so is it true that the > default HZ has changed from 100 to 1000 in RELENG_5? No. FreeBSD dwstable.looksmart.com 5.3-STABLE FreeBSD 5.3-STABLE #73: Mon Feb 14 12:33:38 PST 2005 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/DWTEST i386 dwstable,ttyp0,~,18>sysctl kern.clockrate kern.clockrate: { hz = 100, tick = 1, profhz = 1024, stathz = 128 } -- Doug White| FreeBSD: The Power to Serve [EMAIL PROTECTED] | www.FreeBSD.org ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: HZ in RELENG_5? tcp_subr.c related
On Tue, 22 Feb 2005, Emanuel Strobl wrote: > I read this commit message to RELENG_5: > http://www.freebsd.org/cgi/getmsg.cgi?fetch=436074+438355+/usr/local/www/db/text/2005/cvs-all/20050206.cvs-all > and commented out the HZ=1000 line in my kernel config so I use the > default. But I see only 100 interrupts/sec on clk (with systat) so is > it true that the default HZ has changed from 100 to 1000 in RELENG_5? In RELENG_5, the default HZ for amd64 is 1000, but for all other platforms it is 100. So amd64 users will see a ten-fold decrease in tcp_isn_tick() running, but i386 (and other) users will see no change. Robert N M Watson ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: HZ in RELENG_5? tcp_subr.c related
> Date: Tue, 22 Feb 2005 08:41:58 + (GMT) > From: Robert Watson <[EMAIL PROTECTED]> > Sender: [EMAIL PROTECTED] > > > On Tue, 22 Feb 2005, Emanuel Strobl wrote: > > > I read this commit message to RELENG_5: > > http://www.freebsd.org/cgi/getmsg.cgi?fetch=436074+438355+/usr/local/www/db/text/2005/cvs-all/20050206.cvs-all > > and commented out the HZ=1000 line in my kernel config so I use the > > default. But I see only 100 interrupts/sec on clk (with systat) so is > > it true that the default HZ has changed from 100 to 1000 in RELENG_5? > > In RELENG_5, the default HZ for amd64 is 1000, but for all other platforms > it is 100. So amd64 users will see a ten-fold decrease in tcp_isn_tick() > running, but i386 (and other) users will see no change. Now that 5 is STABLE, I guess we are stuck with it, but in an era of "slow" 2GHz systems, it seems like a questionable choice. I know that there are a lot of folks running old hardware (like my trusty old K6), but they are a minority and changing HZ is not a big deal. Why make the majority live with 100 on fast systems or know enough to manually change it? Has there been discussion of changing this for V6 or is there a reason I missed for keeping HZ at 100 in the iX86 platforms. -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: [EMAIL PROTECTED] Phone: +1 510 486-8634 ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: HZ in RELENG_5? tcp_subr.c related
On Tue, 22 Feb 2005, Kevin Oberman wrote: > > In RELENG_5, the default HZ for amd64 is 1000, but for all other platforms > > it is 100. So amd64 users will see a ten-fold decrease in tcp_isn_tick() > > running, but i386 (and other) users will see no change. > > Now that 5 is STABLE, I guess we are stuck with it, but in an era of > "slow" 2GHz systems, it seems like a questionable choice. I know that > there are a lot of folks running old hardware (like my trusty old K6), > but they are a minority and changing HZ is not a big deal. Why make the > majority live with 100 on fast systems or know enough to manually change > it? > > Has there been discussion of changing this for V6 or is there a reason I > missed for keeping HZ at 100 in the iX86 platforms. In 6-CURRENT, HZ is 1000 for amd64, i386, and ia64, but 100 for other platforms (i.e., ppc, arm, and alpha). I'm not opposed to merging the HZ change to RELENG_5 at some point, but given that occasional nits, such as the TCP nit, are turning up, I think it's worth waiting until after 5.4. Robert N M Watson ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: HZ in RELENG_5? tcp_subr.c related
> Date: Tue, 22 Feb 2005 19:56:03 + (GMT) > From: Robert Watson <[EMAIL PROTECTED]> > > > On Tue, 22 Feb 2005, Kevin Oberman wrote: > > > > In RELENG_5, the default HZ for amd64 is 1000, but for all other platforms > > > it is 100. So amd64 users will see a ten-fold decrease in tcp_isn_tick() > > > running, but i386 (and other) users will see no change. > > > > Now that 5 is STABLE, I guess we are stuck with it, but in an era of > > "slow" 2GHz systems, it seems like a questionable choice. I know that > > there are a lot of folks running old hardware (like my trusty old K6), > > but they are a minority and changing HZ is not a big deal. Why make the > > majority live with 100 on fast systems or know enough to manually change > > it? > > > > Has there been discussion of changing this for V6 or is there a reason I > > missed for keeping HZ at 100 in the iX86 platforms. > > In 6-CURRENT, HZ is 1000 for amd64, i386, and ia64, but 100 for other > platforms (i.e., ppc, arm, and alpha). I'm not opposed to merging the HZ > change to RELENG_5 at some point, but given that occasional nits, such as > the TCP nit, are turning up, I think it's worth waiting until after 5.4. Let's keep STABLE stable. If there are timing issues with HZ of 1000 in V5, we clearly would not want to change before they are resolved. Happily, I have not see these on my faster systems and have not increased HZ on anything < 1 GHz. If I do see network problems, now I know another place to look, too. Thanks! -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: [EMAIL PROTECTED] Phone: +1 510 486-8634 ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: HZ in RELENG_5? tcp_subr.c related
Just put something like this in /boot/loader.conf. kern.hz="250" It works for me since a long time (freebsd 4 I think). Ronald. On Tue, 22 Feb 2005 12:52:06 -0800, Kevin Oberman <[EMAIL PROTECTED]> wrote: Date: Tue, 22 Feb 2005 19:56:03 + (GMT) From: Robert Watson <[EMAIL PROTECTED]> On Tue, 22 Feb 2005, Kevin Oberman wrote: > > In RELENG_5, the default HZ for amd64 is 1000, but for all other platforms > > it is 100. So amd64 users will see a ten-fold decrease in tcp_isn_tick() > > running, but i386 (and other) users will see no change. > > Now that 5 is STABLE, I guess we are stuck with it, but in an era of > "slow" 2GHz systems, it seems like a questionable choice. I know that > there are a lot of folks running old hardware (like my trusty old K6), > but they are a minority and changing HZ is not a big deal. Why make the > majority live with 100 on fast systems or know enough to manually change > it? > > Has there been discussion of changing this for V6 or is there a reason I > missed for keeping HZ at 100 in the iX86 platforms. In 6-CURRENT, HZ is 1000 for amd64, i386, and ia64, but 100 for other platforms (i.e., ppc, arm, and alpha). I'm not opposed to merging the HZ change to RELENG_5 at some point, but given that occasional nits, such as the TCP nit, are turning up, I think it's worth waiting until after 5.4. Let's keep STABLE stable. If there are timing issues with HZ of 1000 in V5, we clearly would not want to change before they are resolved. Happily, I have not see these on my faster systems and have not increased HZ on anything < 1 GHz. If I do see network problems, now I know another place to look, too. Thanks! -- Ronald Klop Amsterdam, The Netherlands ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: HZ in RELENG_5? tcp_subr.c related
Robert Watson <[EMAIL PROTECTED]> writes: > In 6-CURRENT, HZ is 1000 for amd64, i386, and ia64, but 100 for other > platforms (i.e., ppc, arm, and alpha). HZ is (and always was) 1024 on alpha. DES -- Dag-Erling Smørgrav - [EMAIL PROTECTED] ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[EMAIL PROTECTED]"