Which version of the ixgbe driver is included in 5.6? I went through all the RHEL5 release notes, but they only ever state that the driver was updated, not which devices was added.
This device is not supported up to CentOS 5.5 (2.0.44-k2), but the latest driver from the Intel web site works fine (3.2.10).
05:00.0 Ethernet controller: Intel Corporation Unknown device 151c (rev 01)
05:00.0 0200: 8086:151c (rev 01) Subsystem: 8086:a03c Flags: bus master, fast devsel, latency 0, IRQ 170 Memory at de400000 (64-bit, non-prefetchable) [size=2M] I/O ports at ecc0 [size=32] Memory at de3f8000 (64-bit, non-prefetchable) [size=16K] Expansion ROM at de200000 [disabled] [size=1M] Capabilities: [40] Power Management version 3 Capabilities: [50] Message Signalled Interrupts: 64bit+ Queue=0/0 Enable- Capabilities: [70] MSI-X: Enable+ Mask- TabSize=64 Capabilities: [a0] Express Endpoint IRQ 0 Capabilities: [e0] Vital Product Data Capabilities: [100] Advanced Error Reporting Capabilities: [140] Device Serial Number 2c-cf-9a-ff-ff-21-1b-00 Capabilities: [150] Unknown (14) Capabilities: [160] Unknown (16)
On 04/06/2011 05:04 PM, Lars Hecking wrote:
Which version of the ixgbe driver is included in 5.6? I went through all the RHEL5 release notes, but they only ever state that the driver was updated, not which devices was added.
In about 4 hrs time you should be able to check for yourself :)
- KB
Which version of the ixgbe driver is included in 5.6? I went through all the RHEL5 release notes, but they only ever state that the driver was updated, not which devices was added.
This device is not supported up to CentOS 5.5 (2.0.44-k2), but the latest driver from the Intel web site works fine (3.2.10).
FYI current devkernel for RHEL5.7 have it at version 3.2.9-k2, so there is hope it's new enough in 5.6 for you.
Simon
05:00.0 Ethernet controller: Intel Corporation Unknown device 151c (rev 01)
05:00.0 0200: 8086:151c (rev 01) Subsystem: 8086:a03c Flags: bus master, fast devsel, latency 0, IRQ 170 Memory at de400000 (64-bit, non-prefetchable) [size=2M] I/O ports at ecc0 [size=32] Memory at de3f8000 (64-bit, non-prefetchable) [size=16K] Expansion ROM at de200000 [disabled] [size=1M] Capabilities: [40] Power Management version 3 Capabilities: [50] Message Signalled Interrupts: 64bit+ Queue=0/0 Enable- Capabilities: [70] MSI-X: Enable+ Mask- TabSize=64 Capabilities: [a0] Express Endpoint IRQ 0 Capabilities: [e0] Vital Product Data Capabilities: [100] Advanced Error Reporting Capabilities: [140] Device Serial Number 2c-cf-9a-ff-ff-21-1b-00 Capabilities: [150] Unknown (14) Capabilities: [160] Unknown (16)
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Lars Hecking wrote:
Which version of the ixgbe driver is included in 5.6? I went through all the RHEL5 release notes, but they only ever state that the driver was updated, not which devices was added.
This device is not supported up to CentOS 5.5 (2.0.44-k2), but the latest driver from the Intel web site works fine (3.2.10).
05:00.0 Ethernet controller: Intel Corporation Unknown device 151c (rev 01)
05:00.0 0200: 8086:151c (rev 01)
Looking at the 5.6 kernel source (2.6.18-238), the file drivers/net/ixgbe/ixgbe_type.h has the line:
#define IXGBE_DEV_ID_82599_T3_LOM 0x151C
which part of the patch:
linux-2.6-net-ixgbe-update-to-upstream-version-2-0-84-k2.patch
James Pearson
Looking at the 5.6 kernel source (2.6.18-238), the file drivers/net/ixgbe/ixgbe_type.h has the line:
#define IXGBE_DEV_ID_82599_T3_LOM 0x151C
which part of the patch:
linux-2.6-net-ixgbe-update-to-upstream-version-2-0-84-k2.patch
I'll need to test it once available, but may prefer to stick to the newer Intel driver (until 5.7 at least :)
Thanks!
On Wed, Apr 6, 2011 at 9:26 AM, Lars Hecking lhecking@users.sourceforge.net wrote:
Looking at the 5.6 kernel source (2.6.18-238), the file drivers/net/ixgbe/ixgbe_type.h has the line:
#define IXGBE_DEV_ID_82599_T3_LOM 0x151C
which part of the patch:
linux-2.6-net-ixgbe-update-to-upstream-version-2-0-84-k2.patch
I'll need to test it once available, but may prefer to stick to the newer Intel driver (until 5.7 at least :)
I suggest you try asking ELRepo to provide the kmod-ixgbe package built with the Intel version. That would benefit other users with the same hardware. :)
Akemi
On Wednesday, April 06, 2011 06:04:15 pm Lars Hecking wrote:
Which version of the ixgbe driver is included in 5.6? I went through all the RHEL5 release notes, but they only ever state that the driver was updated, not which devices was added.
This device is not supported up to CentOS 5.5 (2.0.44-k2), but the latest driver from the Intel web site works fine (3.2.10).
05:00.0 Ethernet controller: Intel Corporation Unknown device 151c (rev 01)
This does not mean it's not supported it's just not in the pci-id database. You could try to update just the pci-db with: # update-pciids
We've used 82599 on the normal 5.5 kernel no problems (2.0.44-k2). With the 5.6 kernel (-238) you get: 2.0.84-k2
Cheers, Peter
05:00.0 0200: 8086:151c (rev 01) Subsystem: 8086:a03c Flags: bus master, fast devsel, latency 0, IRQ 170
...
This does not mean it's not supported it's just not in the pci-id database. You could try to update just the pci-db with: # update-pciids
Cool!
05:00.0 Ethernet controller: Intel Corporation 82599EB 10 Gigabit TN Network Connection (rev 01)
We've used 82599 on the normal 5.5 kernel no problems (2.0.44-k2). With the 5.6 kernel (-238) you get: 2.0.84-k2
Does not seem to work with our variant. CentOS driver:
ixgbe: Intel(R) 10 Gigabit PCI Express Network Driver - version 2.0.44-k2 ixgbe: Copyright (c) 1999-2010 Intel Corporation.
Intel driver:
Intel(R) 10 Gigabit PCI Express Network Driver - version 3.2.10-NAPI Copyright (c) 1999-2010 Intel Corporation. GSI 25 sharing vector 0xAA and IRQ 25 ACPI: PCI Interrupt 0000:05:00.0[B] -> GSI 39 (level, low) -> IRQ 170 PCI: Setting latency timer of device 0000:05:00.0 to 64 ixgbe: 0000:05:00.0: ixgbe_init_interrupt_scheme: Multiqueue Enabled: Rx Queue count = 16, Tx Queue count = 1 ixgbe: eth2: ixgbe_probe: No DCA provider found. Please start ioatdma for DCA functionality. ixgbe: eth2: ixgbe_probe: (PCI Express:5.0Gb/s:Width x4) 00:1b:21:9a:cf:2c ixgbe: eth2: ixgbe_probe: MAC: 2, PHY: 2, PBA No: G21371-001 ixgbe: eth2: ixgbe_probe: GRO is enabled ixgbe: eth2: ixgbe_probe: HW RSC is enabled ixgbe: eth2: ixgbe_probe: Intel(R) 10 Gigabit Network Connection
On Wednesday, April 06, 2011 07:07:59 pm Lars Hecking wrote:
This does not mean it's not supported it's just not in the pci-id database.
You could try to update just the pci-db with: # update-pciids
Cool!
05:00.0 Ethernet controller: Intel Corporation 82599EB 10 Gigabit TN Network Connection (rev 01)
We've used 82599 on the normal 5.5 kernel no problems (2.0.44-k2). With the 5.6 kernel (-238) you get: 2.0.84-k2
Does not seem to work with our variant. CentOS driver:
ixgbe: Intel(R) 10 Gigabit PCI Express Network Driver - version 2.0.44-k2 ixgbe: Copyright (c) 1999-2010 Intel Corporation.
There are more than one NIC-model with 82599EB and evidently not all work with the CentOS driver. What we're using successfully is X520-DA2:
Intel Corporation 82599EB 10-Gigabit Network Connection (rev 01)
What do you have?
/Peter
There are more than one NIC-model with 82599EB and evidently not all work with the CentOS driver. What we're using successfully is X520-DA2:
Intel Corporation 82599EB 10-Gigabit Network Connection (rev 01)
What do you have?
http://www.intel.com/support/network/sb/cs-012904.htm
PCI id 151c = X520-T2.
Lars Hecking writes:
There are more than one NIC-model with 82599EB and evidently not all work with the CentOS driver. What we're using successfully is X520-DA2:
Intel Corporation 82599EB 10-Gigabit Network Connection (rev 01)
What do you have?
http://www.intel.com/support/network/sb/cs-012904.htm
PCI id 151c = X520-T2.
Just to close this old thread: in a few months of operation, we have repeatedly run into problems where the servers just dropped off the network with pretty much nothing logged on the server or the switch side. The symptoms are strikingly similar to the problem described here (Peter K, you should have remembered :)
http://lists.us.dell.com/pipermail/linux-poweredge/2011-March/044417.html
and the same solution works for us - replacing the standard CentOS ixgbe driver with the latest from Intel.
On Friday, August 05, 2011 04:28:20 PM Lars Hecking wrote:
Lars Hecking writes:
There are more than one NIC-model with 82599EB and evidently not all work with
the CentOS driver. What we're using successfully is X520-DA2: Intel Corporation 82599EB 10-Gigabit Network Connection (rev 01)
What do you have?
http://www.intel.com/support/network/sb/cs-012904.htm
PCI id 151c = X520-T2.
Just to close this old thread: in a few months of operation, we have repeatedly run into problems where the servers just dropped off the network with pretty much nothing logged on the server or the switch side. The symptoms are strikingly similar to the problem described here (Peter K, you should have remembered :)
I guess it slipped my mind since it wasn't me having the problem, heh.
http://lists.us.dell.com/pipermail/linux-poweredge/2011-March/044417.html
and the same solution works for us - replacing the standard CentOS ixgbe driver with the latest from Intel.
Our X520 are still stable except for one recent problem, 2.6.18-238.9.1 -> 2.6.18-238.12.1 broke it quite bad. With 238.12.1 our servers start dropping all incomming packets after a while. Sanity can be (temporarily restored with a "ethtool --negotiate ethX").
We are currently running a kernel that excludes: linux-2.6-net-ixgbe-fix-for-82599-erratum-on-header-splitting.patch since that's what seems to break our setup.
/Peter
2011/8/5 Peter Kjellström cap@nsc.liu.se:
Our X520 are still stable except for one recent problem, 2.6.18-238.9.1 -> 2.6.18-238.12.1 broke it quite bad. With 238.12.1 our servers start dropping all incomming packets after a while. Sanity can be (temporarily restored with a "ethtool --negotiate ethX").
We are currently running a kernel that excludes: linux-2.6-net-ixgbe-fix-for-82599-erratum-on-header-splitting.patch since that's what seems to break our setup.
/Peter
Bad news. Starting with EL5.7, the kernel does not have individual patches any more just like EL6. So, removing a patch is not trivial. You'd have to add a patch that reverts the change. It's a pain but doable...
Akemi
On 8/5/2011 9:46 AM, Peter Kjellström wrote:
Our X520 are still stable except for one recent problem, 2.6.18-238.9.1 -> 2.6.18-238.12.1 broke it quite bad. With 238.12.1 our servers start dropping all incomming packets after a while. Sanity can be (temporarily restored with a "ethtool --negotiate ethX").
We are currently running a kernel that excludes: linux-2.6-net-ixgbe-fix-for-82599-erratum-on-header-splitting.patch since that's what seems to break our setup.
Has the bug been reported upstream?
On Friday, August 05, 2011 05:25:13 PM Les Mikesell wrote:
On 8/5/2011 9:46 AM, Peter Kjellström wrote:
Our X520 are still stable except for one recent problem, 2.6.18-238.9.1 -> 2.6.18-238.12.1 broke it quite bad. With 238.12.1 our servers start dropping all incomming packets after a while. Sanity can be (temporarily restored with a "ethtool --negotiate ethX").
We are currently running a kernel that excludes: linux-2.6-net-ixgbe-fix-for-82599-erratum-on-header-splitting.patch
since that's what seems to break our setup.
Has the bug been reported upstream?
Nope, due to: 1) no reproducer and 2) the "next" EL kernel will have a new major version of the driver code making it quite pointless to invest effort/time in the 2.x driver.
If more information surfaces (maybe in this thread) and/or the problem persists with the 3.x driver then a bz will most likely be created.
/Peter
On Wednesday, August 10, 2011 12:11:51 PM Peter Kjellström wrote:
On Friday, August 05, 2011 05:25:13 PM Les Mikesell wrote:
On 8/5/2011 9:46 AM, Peter Kjellström wrote:
Our X520 are still stable except for one recent problem, 2.6.18-238.9.1 -> 2.6.18-238.12.1 broke it quite bad. With 238.12.1 our servers start dropping all incomming packets after a while. Sanity can be (temporarily restored with a "ethtool --negotiate ethX").
We are currently running a kernel that excludes: linux-2.6-net-ixgbe-fix-for-82599-erratum-on-header-splitting.patch
since that's what seems to break our setup.
Has the bug been reported upstream?
Nope, due to: 1) no reproducer and 2) the "next" EL kernel will have a new major version of the driver code making it quite pointless to invest effort/time in the 2.x driver.
If more information surfaces (maybe in this thread) and/or the problem persists with the 3.x driver then a bz will most likely be created.
To follow up on this, we can not reproduce the problem with the 3.x driver as shipped with the 5.7 kernel.
/Peter
-
Akemi Yagi -
James Pearson -
Karanbir Singh -
Lars Hecking -
Les Mikesell -
Peter Kjellström -
Simon Matter