[syslinux] gpxelinux.0 and slow HTTP performance on VMware ESX VM

Schlomo Schapiro syslinux at schlomo.schapiro.org
Thu Jun 30 22:15:04 PDT 2011


Hi,

On 30/06/11 22:08, Eric W. Biederman wrote:
> It is odd, because at least the pcap shows that 4.10-pre15 completed
> the http transfer of /boot/pxelinx.cfg/default and went on to try and=
> transfer the next file.  In trying to establish the next tcp connect
> pxelinux decided to RST the connection instead of just acking it.
> 
> And there it looks like there was a little bit of confusing in the final
> handshake of the transfer of /boot/pxelinux.cfg/default
> 
> What machine was capturing the packet trace?
> I assume you were working against apache given the access log.  Was the
> apache server located on another VM?

the trace was done on the boot server where also the apache and tftp server
runs. This is another VM on the same ESX server.

If it helps I could post another trace of gpxelinux.0 4.04 which looks much worse.

I am also unclear as to why pxelinux 4.10 would not load vesamenu.c32 (UI
VESAMENU.C32) but try to load vesamenu.c32.0 and vesamenu.c32.com before giving up.

Kind Regards,
Schlomo




More information about the Syslinux mailing list