From: Brian Capouch Date: Mon, 02 Dec 2002 01:45:18 +0000 Subject: [LARTC] How accurate should HTB be? Message-Id: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: lartc@vger.kernel.org I have been playing with HTB for a couple of weeks on one of my testbed routers, and was astounded at how accurate it seemed to be (measuring with iperf). So I rolled it out on two production machines, and now, Murphy's Law, those both seem to be throttling too much. Same hardware, same kernel version, same iproute2 utils. I have pored over the output of tc -s "shows" and the only thing I can see different between the "highly accurate" machine and the two that are "overthrottled" is shown in the snippets below. I'm grasping at straws; does it matter that the rate is being reported differently by each invocation of tc? The upload speed of the first runs 252, 258, 254, etc.; on the second 86, 150, 92, 78, etc. I don't know how to debug this, and hope this isn't something obvious and stupid. Thanks in advance for any help that might be forthcoming. B. *********************************************** Good: $ tc -s -d class show dev eth0 class htb 1:1 root rate 256Kbit ceil 256Kbit burst 1926b/8 mpu 0b cburst 1926b/8 mpu 0b level 7 Sent 18417658 bytes 132268 pkts (dropped 0, overlimits 0) rate 2bps lended: 0 borrowed: 0 giants: 0 tokens: 45375 ctokens: 45375 Overthrottled: $ tc -s -d class show dev eth1 class htb 1:1 root rate 256Kbit ceil 256Kbit burst 1926b/8 mpu 0b cburst 1926b/8 mpu 0b level 7 Sent 6612505 bytes 10225 pkts (dropped 0, overlimits 0) rate 6179bps 4pps lended: 0 borrowed: 0 giants: 0 tokens: 45375 ctokens: 45375 ************************************************* _______________________________________________ LARTC mailing list / LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/