[LARTC] Re: What am I missing?
Follow-up to my own post here: http://marc.theaimsgroup.com/?l=lartc&m=115332794424197&w=2 Problem ended up being that the tc included in the iproute RPM in Fedora Core 2 was built specifically against the headers in the glibc-kernheaders RPM. When I switched to a custom kernel (for built-in MPPE support), apparently some symbols referenced in the glibc-kernheaders package are not present in the non-RH kernel. This was causing my issue. Solution was to modify the .spec file for iproute and tell it to compile against my custom kernel's headers specifically. Specifically I had to make the following changes to the .spec file to get things to work: - Modify iproute2-2.4.7-kernel.patch to point to /usr/src/linux/include - Do not apply iproute2-2.4.7-misc.patch - Do not apply iproute2-2.4.7-in_port_t.patch This makes everything happy once again! I'm not sure what the consequenes are of not applying those RedHat patches, but the tool still works normally as far as I can tell. Ray ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
[LARTC] What am I missing?
On a custom compiled Linux 2.6.13 kernel... # tc qdisc add dev ppp145 handle : ingress # tc filter add dev ppp145 parent : protocol ip prio 50 u32 match ip src 0.0.0.0/0 police rate 384kbit burst 10k drop flowid :1 RTNETLINK answers: Invalid argument This works fine on a CentOS machine and my Fedora Core 2 box with default kernel. I'm trying to figure out what is missing in the custom built 2.6.13 kernel. strace isn't real helpful, and there doesn't appear to be a way to make tc present more useful error messages... Non ingress filtering works. Here's an excerpt from kernel config: # # QoS and/or fair queueing # CONFIG_NET_SCHED=y CONFIG_NET_SCH_CLK_JIFFIES=y # CONFIG_NET_SCH_CLK_GETTIMEOFDAY is not set # CONFIG_NET_SCH_CLK_CPU is not set CONFIG_NET_SCH_CBQ=y CONFIG_NET_SCH_HTB=y CONFIG_NET_SCH_HFSC=y CONFIG_NET_SCH_PRIO=y CONFIG_NET_SCH_RED=y CONFIG_NET_SCH_SFQ=y CONFIG_NET_SCH_TEQL=y CONFIG_NET_SCH_TBF=y CONFIG_NET_SCH_GRED=y CONFIG_NET_SCH_DSMARK=y CONFIG_NET_SCH_NETEM=y CONFIG_NET_SCH_INGRESS=y CONFIG_NET_QOS=y CONFIG_NET_ESTIMATOR=y CONFIG_NET_CLS=y # CONFIG_NET_CLS_BASIC is not set CONFIG_NET_CLS_TCINDEX=y CONFIG_NET_CLS_ROUTE4=y CONFIG_NET_CLS_ROUTE=y CONFIG_NET_CLS_FW=y CONFIG_NET_CLS_U32=y CONFIG_CLS_U32_PERF=y CONFIG_NET_CLS_IND=y # CONFIG_CLS_U32_MARK is not set CONFIG_NET_CLS_RSVP=y CONFIG_NET_CLS_RSVP6=y # CONFIG_NET_EMATCH is not set # CONFIG_NET_CLS_ACT is not set CONFIG_NET_CLS_POLICE=y Any ideas? Can paste entire kernel config if necessary. ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
[LARTC] Egress shaping recommendations (based on source ip)
I'm trying to set up customer egress shaping on concentrator that customers connect to using pptp (which uses pppd). This means customers each have a pppXXX interface when they connect. It's easy to shape their "down" speed this way, and when they disconnect the interface is gone and thus the shaping rules automatically go away as well. However, shaping their upload (traffic that forwards from their ppp interface and out the ethernet interface on the server) is a different matter and I'm trying to figure out the best way to do it. I was planning to do this with CBQ... initially I created a root qdisc with the full Ethernet bandwidth available. I created two classes below that, one with the full bandwidth available, and one with only 9Mbps available (internet bandwidth we have at this site). To the 9Mbps class, I want to attach child classes limited to the rate plan each customer is assigned. These classes should be created when the customer connects, and removed when they disconnect. I will create a filter based on "ip src" and point to the appropriate class for each customer. The problem I'm running into is that when I customer disconnects, I cannot find a "clean" way to remove their associated classes and filters. I would like to be able to do simply delete the class I created for them and have all the attached filters automatically removed as well... kind of like how you can delete the root qdisc and everything below is removed automatically. However, this doesn't seem to work. You have to delete the filters first before removing any classes. Instead of using classes should I be looking at using child qdisc's and attaching filters directly to those (if that can be done)? Just trying to avoid a situation on a busy server where a tc command fails and I'm left with stale filters and classes. Ray ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
