Date: Tue, 13 Mar 2012 09:19:05 -0000 From: "Steven Hartland" <killing@multiplay.co.uk> To: "Jack Vogel" <jfvogel@gmail.com> Cc: freebsd-net@freebsd.org, =?KOI8-R?B?68/O2MvP1yDl18fFzsnK?= <kes-kes@yandex.ru> Subject: Re: Re[2]: ixgbe interface micro stalls / slow responses Message-ID: <32B098EB44ED44E2B9724D014F44A8A2@multiplay.co.uk> References: <18969B34B4EE402986389D5022DCCB35@multiplay.co.uk><853549959.20120312184940@yandex.ru><2B4507129C2141BC9D12B6F7BE5987B3@multiplay.co.uk><50980382.20120312202136@yandex.ru><86016A34935443C383B3EE08F3D4292D@multiplay.co.uk><CAFOYbckT_xsbR1Qdtf%2BNscFaRedPqjLhtRzcsJ%2B%2B3F0XfYaqfA@mail.gmail.com><2ABAFC6511784968A02D9580D8EB2AE7@multiplay.co.uk><CAFOYbc=MvUDxQwWQdKKwO9PVU6FO%2BTuYKiUpwR8kZOJFxs2Zqw@mail.gmail.com><C3CD71686C1047FFBD89691D9B8C176F@multiplay.co.uk> <CAFOYbcnyhB-vTsm_Umgg4Cgk_Rdw2UgKgc_BdPsybpb-TScHsg@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Same behaviour with just one queue:-
Packets Pings
Host Loss% Snt Last Avg Best Wrst StDev
1. ixtest 0.0% 28 0.1 24.4 0.1 297.5 73.0
ix0: <Intel(R) PRO/10GbE PCI-Express Network Driver, Version - 2.4.5> port 0x2000-0x201f mem 0xd8400000-0xd847ffff,0xd8480000-0xd8483fff irq 52 at device 0.0 on pci5
ix0: Using MSIX interrupts with 2 vectors
ix0: [ITHREAD]
ix0: [ITHREAD]
ix0: Ethernet address: 00:1b:21:7e:2e:8c
ix0: PCI Express Bus: Speed 5.0Gb/s Width x8
Setting the following in loader.conf does fix the RX warning, so could do with updating the docs so it lists /boot/loader.conf instead of /etc/sysctl.conf
kern.ipc.nmbjumbop=3D262144
kern.ipc.nmbclusters=3D524288
Regards
Steve
----- Original Message -----=20
From: Jack Vogel=20
To: Steven Hartland=20
Cc: =EB=CF=CE=D8=CB=CF=D7 =E5=D7=C7=C5=CE=C9=CA ; freebsd-net@freebsd.org=20
Sent: Monday, March 12, 2012 8:32 PM
Subject: Re: Re[2]: ixgbe interface micro stalls / slow responses
I don't know what to make of it right now, i would suggest going minimal, meaning go to a single
queue and see what the behavior is.=20
Jack
2012/3/12 Steven Hartland <killing@multiplay.co.uk>
Looks like nmbclusters sysctl changes need to loader.conf so they are available early enough for the driver.
We haven't rebooted to test that yet as the machine is under so little network load I wouldn't expect raising it from 1024 to 2048 RX descriptors to make any real difference, what do you recon?
=20
Regards
Steve
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
This e.mail is private and confidential between Multiplay (UK) Ltd. and the person or entity to whom it is addressed. In the event of misdirection, the recipient is prohibited from using, copying, printing or otherwise disseminating it or any information contained in it.=20
In the event of misdirection, illegible or incomplete transmission please telephone +44 845 868 1337
or return the E.mail to postmaster@multiplay.co.uk.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?32B098EB44ED44E2B9724D014F44A8A2>