Is there a reason you're not using PV? The RealTek emulated NIC is buggy and horrible, and I wish XenServer/XCP would let you change it like normal Xen. I believe Citrix's stance was that they only supported HVM (Windows) VMs when running their PV drivers so that the issue was moot.
You might be able to hack the code you let you use e1000, which might work, but you're really better off using PV.
On May 18, 2011 1:36 PM, "Gregory Durham" <gregory.durham@xxxxxxxxx
> Hello all,
> Not sure if I am completely missing something but here goes. I have
> XCP 1.0 running without any issues. I have the mtu set on both the
> storage network, the corresponding PIF, and the VIF to 9000 and when
> displayed in ifconfig it shows that this is indeed true. I have
> several CentOs 5.6 VMs running in HVM mode (i.e. no xen kernel) which
> attach to the the storage network through these VIFs and when I
> attempt to enable jumbo frames, I receive: SIOCSIFMTU: Invalid
> argument, I receive the same error from ifconfig eth1 mtu 9000 and
> making the change in the ifcfg-eth1 config file. This is the driver I
> have running from lspci -vv
> 00:04.0 Ethernet controller: Realtek Semiconductor Co., Ltd.
> RTL-8139/8139C/8139C+ (rev 20)
> Subsystem: XenSource, Inc. Device 0001
> Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr-
> Stepping- SERR- FastB2B- DisINTx-
> Status: Cap- 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort-
> <TAbort- <MAbort- >SERR- <PERR- INTx-
> Latency: 64, Cache Line Size: 32 bytes
> Interrupt: pin A routed to IRQ 177
> Region 0: I/O ports at c100 [size=256]
> Region 1: Memory at f3001000 (32-bit, non-prefetchable) [size=256]
> Kernel driver in use: 8139cp
> Kernel modules: 8139too, 8139cp
> If I am doing something wrong, please let me know. Or, if it is just
> plain not possible. I am pulling my hair out.
> Xen-users mailing list