So far I tested PXELINUX from 4.10-pre12, 4.10-pre9 and 4.10-pre6. The first two take a long time with a TFTP file not found and eventually find my file but error out (presumably to kaboom). The last takes a long time, does a lot of retransmitted packets (didn't watch the first two), then eventually loads the file but continues this behavior for TFTP and HTTP transfers. This is on VMware Server 2.0.2-203138 in Linux using the DHCP server that comes with VMware Server, the PXE ROM from VMware Server (info below) and tftpd-hpa 5.0. HPA, if there's a test you'd like me to try (patch, different -pre, etc) let me know. PXE ROM strings: Network boot from AMD Am79c970A Copyright (C) 2003-2005 VMware, INC. Copyright (C) 1997-2000 Intel Corporation UNDI code segment at 9ECC len 0A0D UNDI data segment at 9C7F len 24D0 -- -Gene
On Tue, May 3, 2011 at 22:07, Gene Cumm <gene.cumm at gmail.com> wrote:> So far I tested PXELINUX from 4.10-pre12, 4.10-pre9 and 4.10-pre6. > The first two take a long time with a TFTP file not found and > eventually find my file but error out (presumably to kaboom). ?The > last takes a long time, does a lot of retransmitted packets (didn't > watch the first two), then eventually loads the file but continues > this behavior for TFTP and HTTP transfers. ?This is on VMware Server > 2.0.2-203138 in Linux using the DHCP server that comes with VMware > Server, the PXE ROM from VMware Server (info below) and tftpd-hpa 5.0. > > HPA, if there's a test you'd like me to try (patch, different -pre, > etc) let me know. > > PXE ROM strings: > > Network boot from AMD Am79c970A > Copyright (C) 2003-2005 ?VMware, INC. > Copyright (C) 1997-2000 ?Intel Corporation > > > UNDI code segment at 9ECC len 0A0D > UNDI data segment at 9C7F len 24D0I forgot to mention the obvious: This setup works with 4.04. -- -Gene
On 05/03/2011 07:07 PM, Gene Cumm wrote:> So far I tested PXELINUX from 4.10-pre12, 4.10-pre9 and 4.10-pre6. > The first two take a long time with a TFTP file not found and > eventually find my file but error out (presumably to kaboom). The > last takes a long time, does a lot of retransmitted packets (didn't > watch the first two), then eventually loads the file but continues > this behavior for TFTP and HTTP transfers. This is on VMware Server > 2.0.2-203138 in Linux using the DHCP server that comes with VMware > Server, the PXE ROM from VMware Server (info below) and tftpd-hpa 5.0. > > HPA, if there's a test you'd like me to try (patch, different -pre, > etc) let me know. > > PXE ROM strings: > > Network boot from AMD Am79c970A > Copyright (C) 2003-2005 VMware, INC. > Copyright (C) 1997-2000 Intel Corporation > > > UNDI code segment at 9ECC len 0A0D > UNDI data segment at 9C7F len 24D0 >I know the VMware PXE stack is awful beyond belief. However, could you get a packet trace (e.g. wireshark) by any chance? -hpa -- H. Peter Anvin, Intel Open Source Technology Center I work for Intel. I don't speak on their behalf.
Gene Cumm wrote:> So far I tested PXELINUX from 4.10-pre12, 4.10-pre9 and 4.10-pre6. > The first two take a long time with a TFTP file not found and > eventually find my file but error out (presumably to kaboom). The > last takes a long time, does a lot of retransmitted packets (didn't > watch the first two), then eventually loads the file but continues > this behavior for TFTP and HTTP transfers. This is on VMware Server > 2.0.2-203138 in Linux using the DHCP server that comes with VMware > Server, the PXE ROM from VMware Server (info below) and tftpd-hpa 5.0. > > HPA, if there's a test you'd like me to try (patch, different -pre, > etc) let me know. > > PXE ROM strings: > > Network boot from AMD Am79c970A > Copyright (C) 2003-2005 VMware, INC. > Copyright (C) 1997-2000 Intel Corporation > > > UNDI code segment at 9ECC len 0A0D > UNDI data segment at 9C7F len 24D04.10-pre1 works in my VMware Player 3.0.0; 4.10-pre2 doesn't work anymore (really slow config file loading). Sebastian
Hi, I am seeing very slow startup times for the 4.10-pre14 version. The 4.04/4.05-pre1 work just fine. What happens is that there is are extremely large delays (minutes) when the TFTP server and client are going through all the "not available' files until they reach the 'pxelinux.cfg/default' file. (JFYI, The booting has not reached sending the vmlinuz yet. Unlike some other folks who are seeing slowness errors in the HTTP part) This TFTP delay is seen on real hardware (Dell Poweredge 2850) or VMWare ESX based VMs. Same for VirtualBox VMs I am using the e1000 chipset for the LAN for VMWare ESX and VirtualBox. lspci for the Dell PowerEdge 2850: # lspci | grep -i eth 06:07.0 Ethernet controller: Intel Corporation 82541GI Gigabit Ethernet Controller (rev 05) 07:08.0 Ethernet controller: Intel Corporation 82541GI Gigabit Ethernet Controller (rev 05) 0a:02.0 Ethernet controller: Intel Corporation 82545GM Gigabit Ethernet Controller (rev 04) 0a:03.0 Ethernet controller: Intel Corporation 82545GM Gigabit Ethernet Controller (rev 04) Regards, Shantanu> H. Peter Anvinhpa at zytor.com > Sun May 8 18:56:06 PDT 2011________________________________> I just made a couple of changes... would you mind re-running your vmware > >tests? >
On 05/26/2011 10:33 PM, Shantanu Gadgil wrote:> > This TFTP delay is seen on real hardware (Dell Poweredge 2850) or VMWare ESX > based VMs. > Same for VirtualBox VMs >Any way you could obtain a packet trace (using tcpdump or wireshark)? -hpa