Re: [pfSense] High interrupt load on LAGG with LACP
Hi, you where right. It was/is kind of a traffic problem. I just did the test with one cable and the interrupt load was already at 60-80%. Twice that will of course kill the box. Thanks for the fast help. CU Jens ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
Are you sure it's not a networking problem ? When the traffic stops after a few minutes might it be related to stp or else ? Are the connected switches lacp compliant ? And configured ? Am 05.06.2012 um 17:44 schrieb "Jens Kühnel" : > Am 05.06.12 00:33, schrieb Chris Buechler: > Hi, > thanks for the fast responds. > >> Ultimately your problem is trying to increase bandwidth on a box that >> can only push 85 Mbps. When you get that to a level that it can >> potentially have 200 Mbps flowing to/through it, you're going to >> drag it down considerably. You just need a faster box, an ALIX isn't >> a good fit for that kind of scenario. > > Sorry, I forgot to mention. The interrupt load happens when no traffic > is going through. It's not a traffic problem. > > I just tried it and the traffic rate is increased (about 15MB/s) when > both cables are plugged in, but only for a couple of minutes. Than the > the interrupt load starts rising above 90% and the system gets slower. > > CU > Jens > > ___ > List mailing list > List@lists.pfsense.org > http://lists.pfsense.org/mailman/listinfo/list > ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
On Tue, Jun 5, 2012 at 9:42 AM, Jens Kühnel wrote: > Sorry, I forgot to mention. The interrupt load happens when no traffic > is going through. It's not a traffic problem. What you're describing is odd, but I will echo Chris' point, which is that the ALIX has no hope whatsoever of routing at speeds that will saturate a 100 mbps connection, so you have absolutely nothing to gain by using LAGG on it. That's not to say that you shouldn't troubleshoot it, but you should realize that the pursuit is purely academic. db ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
Am 05.06.12 00:33, schrieb Chris Buechler: Hi, thanks for the fast responds. > Ultimately your problem is trying to increase bandwidth on a box that > can only push 85 Mbps. When you get that to a level that it can > potentially have 200 Mbps flowing to/through it, you're going to > drag it down considerably. You just need a faster box, an ALIX isn't > a good fit for that kind of scenario. Sorry, I forgot to mention. The interrupt load happens when no traffic is going through. It's not a traffic problem. I just tried it and the traffic rate is increased (about 15MB/s) when both cables are plugged in, but only for a couple of minutes. Than the the interrupt load starts rising above 90% and the system gets slower. CU Jens ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
Op 5-6-2012 3:53, Glenn Kelley schreef: Good to know. For us we just need 100-300mbps in the sky (literally 300 foot up a tower) The soekris net6501 may be a good fit, it can do PoE iirc. It's a 600-1.6Ghz Intel Atom. I've benchmarked the faster Intel Atom 1.8 Dual core in a Lanner Inc FW7535 at 220mbit full duplex. Cheers, Seth ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
Good to know. For us we just need 100-300mbps in the sky (literally 300 foot up a tower) Blessings Joe Landman wrote: e boxen for customers for pfSense with up to 8x 1GbE ports, and several with dual 10GbE ports. Building the 10GbE driver for 2.0.1 was a bear, and it doesn't load correctly on some of the units*, but these are fairly capable units, and we were pushing about 1Gb/s thro ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
On 06/04/2012 09:38 PM, Glenn Kelley wrote: Chris That is good to know. I have some wireless backhauls pushing well over 100mbps So better to know now vs later any suggestions on hardware for the sky in that case? We've built some boxen for customers for pfSense with up to 8x 1GbE ports, and several with dual 10GbE ports. Building the 10GbE driver for 2.0.1 was a bear, and it doesn't load correctly on some of the units*, but these are fairly capable units, and we were pushing about 1Gb/s through a unit at a customer site. People are inclined to skimp on these designs ... its a mistake if you have lots of traffic to move. * a request for the next pfSense release would be driver building environment installable package (much like the other packages) with the minimum subset of tools we need to compile drivers for pfSense. We can usually fix/port drivers for alternative versions of FreeBSD, but sometimes, its very helpful to have the exact version of the kernel headers/compilers used. -- Joseph Landman, Ph.D Founder and CEO Scalable Informatics Inc. email: land...@scalableinformatics.com web : http://scalableinformatics.com http://scalableinformatics.com/sicluster phone: +1 734 786 8423 x121 fax : +1 866 888 3112 cell : +1 734 612 4615 ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
Chris That is good to know. I have some wireless backhauls pushing well over 100mbps So better to know now vs later any suggestions on hardware for the sky in that case? Chris Buechler wrote: Ultimately your problem is trying to increase bandwidth on a box that can only push 85 Mbps. When you get that to a level that it can potentially have 200 Mbps flowing to/through it, you're going to drag it down considerably. You just need a faster box, an ALIX isn't a good fit for that kind of scenario. ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
Re: [pfSense] High interrupt load on LAGG with LACP
On Mon, Jun 4, 2012 at 2:53 PM, Jens Kühnel wrote: > Hi, > > I'm a longtime and very satisfied pfsense user. But I ran into an > situation that I can not understand. Maybe you can give me a Clue what > to do. Or where to look for infos. > > I have Alix 2D3 Board with PFSense 2.0.1-RELEASE i386 running on my > CF-Card. I switched recently to a new switch. A Cisco SG-300 28 gigabit > ports (the Small Office version without IOS). To increase the bandwidth > I switched from vr0 interface to a LAGG with vr0 and vr2 running in LACP > Mode. I configured the switch to use LAG with LACP Enabled with LACP > Timeout long. I use VLans on top of the vr0 and now on top of the LAG. > > When the system comes up everything looks good. But inside 1-5 Minutes > the CPU gets eaten up from interrupts, up to 99.9% and then the system > freezes. When I remove one of the links, the interrupt load goes down > instendlty, and settles around 10-50% depending on the load of the > interface. > Ultimately your problem is trying to increase bandwidth on a box that can only push 85 Mbps. When you get that to a level that it can potentially have 200 Mbps flowing to/through it, you're going to drag it down considerably. You just need a faster box, an ALIX isn't a good fit for that kind of scenario. ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
[pfSense] High interrupt load on LAGG with LACP
Hi, I'm a longtime and very satisfied pfsense user. But I ran into an situation that I can not understand. Maybe you can give me a Clue what to do. Or where to look for infos. I have Alix 2D3 Board with PFSense 2.0.1-RELEASE i386 running on my CF-Card. I switched recently to a new switch. A Cisco SG-300 28 gigabit ports (the Small Office version without IOS). To increase the bandwidth I switched from vr0 interface to a LAGG with vr0 and vr2 running in LACP Mode. I configured the switch to use LAG with LACP Enabled with LACP Timeout long. I use VLans on top of the vr0 and now on top of the LAG. When the system comes up everything looks good. But inside 1-5 Minutes the CPU gets eaten up from interrupts, up to 99.9% and then the system freezes. When I remove one of the links, the interrupt load goes down instendlty, and settles around 10-50% depending on the load of the interface. I can't enable the debug infos on the switch, there is no option for that. But the same configuration is used with Gigabitports on a RHEL6 Linux system without any Problems. I'm more a Linux kind of guy, so I don't know where to have a look what is using the interrupts. One thing I noticed is that I have Hardware Checksum offloading enabled, but Hardware TCP Segmentation Offloading und HW Large Receive Offloading is disabled under System - Advanced - Networking. I don't remember changing this, but the System was installed on PFSense 1.2 and updated when a new release came out. Is someone using this kind of switch or an Alix board with LAGG and LACP and can share his results? Thanks for this great product and for your help with this. Greeting from Frankfurt Germany Jens ___ List mailing list List@lists.pfsense.org http://lists.pfsense.org/mailman/listinfo/list
