PXE Boot Timeout Issue...

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Hi all,

Just upgraded to 12.3 user space tools from 11.0, and now when I attempt to netboot a guest, it appears that the pxe rom is timing out on dhcp before the bridge has enough time to come up.

Is there a command line switch to set the dhcp timeout, or a build option that can be changed to set the timeout to a longer value, or disable it entirely?

Thanks in advance...

Stu

- --
Spider Pig, Spider Pig, Does whatever a Spider Pig does. Can he swing, from a web, no he can't, he's a pig. Look Oouuuut! He is a Spider Pig....
-- Matt Groening - "Spider Pig Lyrics"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iQIcBAEBCAAGBQJL0O2PAAoJEFKVLITDJSGSHKUP/RGNTgmaQmqQ4L5wdkRpWVY4 0BY0wsfajLCQ8W87Zr5EIoFDw3QfeKM9345bvstwSvzvX5WB19U/Q463nxzUfV3v YjBk6oCQVI7nnbSZIuz0rqxeo8QPcfuDStHQAkYZD6TlfgNntbzFRnCoMRAMuK46 EdcjmJMCcZ5bF2OezzljhpvJavqwBhOsuOwXHprlIxkhnlOQB5sBY4RSS/f886xw 34Bf1ER7J3zoyq96ElTOCNpFU5BeoaUGw0n/fDPaxQ5QmGoHOB/OpSoN8fae+h1J OxgPWWUhgcb1d0rCsU9R4frbzJGwQFy8X4tBw4gqhjPX76D3T4zyf532pbTjrnri GASOLFuSGxowPH1LmPag36Yq8IdOejVrEzK8jTyrpbmWtVBkyS9NiqQEJmqdbGpt bu4Kt+Ab4oJRI+DGf8kfnvBvfZ5wIPmfsWzV3YrV4AHn+9X6FG/1OMxg9EV8cyC7 hhy/BTGX19qwzuxxUq8s1aH0BSARip++lZoKFuYojF5AoP+SBUSoOjpdhY+NNMTh 3cldyrN3Zd9Y5/ejHIBta0QDUg6YdbXvi6Ai2B+6aKtq63B4+uT8XBfh+duKNBfF Y25gERY2aXoed4KRukSfr/6qh2y9MSQb7UbZneAy9bFMKhK21/ZK9FRM/uIUOs+g lE6iqHCL4f4esyNFIwAr
=OAyU
-----END PGP SIGNATURE-----

The bridge shouldn't need significant amounts of time to come up. Can you describe the networking setup? Are you using libvirt and with what network config?

If you have a bridge configured, can you show the output of:

$ sudo brctl showstp $bridge_name

Stefan

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

The network is fairly straight forward. PC with single NIC directly attached to switch with a local DHCP server.

We do not use libvirt, we are using qemu with command switches.

Interestingly it appears that the -option-rom command switch doesn't do anything any longer. It is using gPXE by default only.

Here is our current command line we are testing with:

export QEMU_AUDIO_DRV=alsa

qemu-system-x86_64 \
        -m 1024 \
        -soundhw es1370 \
        -name 'Net Boot Disk' \
        -net nic,macaddr=52:54:00:12:31:06,model=virtio \
        -net tap,ifname=tap6,script=/etc/kvm/kvm-ifup-br0 \
        -option-rom /usr/local/share/qemu/pxe-virtio.bin \
        -boot n \
        -daemonize

Here is the exact chain of events:

1) Start guest from command line
2) tap adapter is added to br0 and enters learning state on host while gPXE starts DHCP request on guest.
3) after about 15 seconds, tap device enters promiscuous mode on host, and at same time, gPXE reports "No bootable device." on guest
4) Drop to guest console and signal a system_reset, guest gets DHCP address and netboots normally.

Here is the info you requested, both before and after guest startup.

Before guest startup:

linus:~# brctl show br0
bridge name	bridge id		STP enabled	interfaces
br0		8000.001cc092ba91	no		eth0

linus:~# brctl showstp br0
br0
 bridge id		8000.001cc092ba91
 designated root	8000.001cc092ba91
 root port		   0			path cost		   0
 max age		  20.00			bridge max age		  20.00
 hello time		   2.00			bridge hello time	   2.00
 forward delay		  15.00			bridge forward delay	  15.00
 ageing time		 300.00
 hello timer		   1.87			tcn timer		   0.00
 topology change timer	   0.00			gc timer		   8.87
 flags			

eth0 (1)
port id 8001 state forwarding
designated root 8000.001cc092ba91 path cost 19
designated bridge 8000.001cc092ba91 message age timer 0.00
designated port 8001 forward delay timer 0.00
designated cost 0 hold timer 0.87
flags

After guest startup:

linus:~# brctl show br0
bridge name	bridge id		STP enabled	interfaces
br0		8000.001cc092ba91	no		eth0
							tap6
linus:~# brctl showstp br0
br0
 bridge id		8000.001cc092ba91
 designated root	8000.001cc092ba91
 root port		   0			path cost		   0
 max age		  20.00			bridge max age		  20.00
 hello time		   2.00			bridge hello time	   2.00
 forward delay		  15.00			bridge forward delay	  15.00
 ageing time		 300.00
 hello timer		   1.00			tcn timer		   0.00
 topology change timer	   0.00			gc timer		   2.00
 flags			

eth0 (1)
port id 8001 state forwarding
designated root 8000.001cc092ba91 path cost 19
designated bridge 8000.001cc092ba91 message age timer 0.00
designated port 8001 forward delay timer 0.00
designated cost 0 hold timer 0.00
flags

tap6 (2)
port id 8002 state forwarding
designated root 8000.001cc092ba91 path cost 100
designated bridge 8000.001cc092ba91 message age timer 0.00
designated port 8002 forward delay timer 0.00
designated cost 0 hold timer 0.00
flags

Let me know if you need anything else...

Stu

- --
Well I've always had a deep respect, And I mean that most sincerely. The band is just fantastic, that is really what I think. Oh by the way, which one's Pink?
-- Pink Floyd - "Have a Cigar Lyrics"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iQIcBAEBCAAGBQJL0cdkAAoJEFKVLITDJSGSRvsQAIErw0N4aCETAxeEq27Z5cgR 0V3jLcHfRurMr0LSXVxJ4hijOCOVblYHaplMHh4M5qFh9DvKg+ZsJrQagUCaxOeT lk3xQBDWQ5g23CTbgVmyaGVCfeiEVBurGvDT/z7X5bs+w4aaSMbQrlEp6KrnehQW xcUj1m6Q1/rnjvVaoXUSQbxmIoIxCn7CWC+bVN2ufXes73rWIK076qRmtbHEzaHr ImxxQVNnOPGXVW/HtOgRhP/lky2kXAgjvCifs2ul5Gs9iiHZp8LepVcaHLpxk9FQ BgvnT0ayTa8AmgzZwIolrRvZ6FHLJ2UHgCVnUh8vFQhBSls0XJZPpOZC50xK04lX Q5vzVLon8klCEDUkrASQFQ3WrBjRoXiqK6Vz2p2aUE3svLP6Z9WAPOj/c17SdLgb iH5Q8qjj1cEpmsrpUkFhOH51MWoYmlJWATI+QSxaa9ksrKO73Nr6WEmPLHtrKm40 ahXKZddbG0hU3Bqp31emEmGLuYowdOwE9DGG5M+6vW0NUNEXBoBf/6Nn9w2uxKj9 oYxMcAQt7yKgvJl7QNxs6pGIHHnay59jUPePyeu235EwYNBYjXIc2I8wI9+ua42j yTszPnh+NLrCpmNfhcSQJN8NeNp9c6J11Nct+RJZ8TkQjiY9X2dWc/bU/65naoSP s49qoa4OQdBRxvkyR52n
=YLw4
-----END PGP SIGNATURE-----

....

...

I've run into this issue as well and lowering the forwarding delay with "brctl setfd br0 5" fixes it.

-jim

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Thanks Jim! I think that has it...

Stu

- --
Hey nineteen Thats retha franklin, She dont remember The queen of soul, Its hard times befallen The sole survivors, She thinks Im crazy, But Im just growing old
-- Steely Dan - "Hey 19 - Lyrics"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iQIcBAEBCAAGBQJL0dkFAAoJEFKVLITDJSGSwJAQAIBL16jFmTavFWX0dU50syP+ VtK3EqQcv02bJdpAXdbfKuO1QDdjmucG2zEbNK4lMXKEwBsGcKtxDhUOVaNS07fK OdfvsO+/wX9mC8JmqCah6r8Pwppd3F07nzP3HkPnswV/rgtDVrSWB2JInDy+d3R7 sd0DisPOeaHoXSQqJ8REGElVR2adzsdmRWUKpFmtu9tJCz48wsbkDLpcr5iMt7jA sAPtTm1ENUEF3X3l6mlGB4kuYsLum93hPpgrxWmMvGO3uokVzRJYm3x2Yf5J++3v yc8reMPPsQe1tI9r+ZtCKXE2XLqI/X4Pdap0IIjtid2245dBOMKys+aIXYZ1d7jL YwngcgFDdse6B3Gt9/Tk5RhqqYQeYLsZckMOhy89DN2bftYxsX70xIWKKVxPEkNg GniSN8pCg/Rx4ZOFkygwXRUayOx5Hi5nR6hetKyBqRM6mWYZvfvzRYH3O05JAOIz 3qKRObyuKHYFGkdWZ1LaTBoQoOdq+iWyuPK53kvpHFbdks6mZZg3YO70M6+ziqyT zy9UvuumjzquuFwRu91R0qi02SZlNifvTQZ9aUpsPvRAhijroidqIzsH/5CoDuDe AyQEKJgq64/ViA2QrYgpo/IFLAUbhCpMRIr7Tt5Te2TT+s43Lefyg6aY2XA2UwY0 6LwIWdBWwITjvP7XUU5B
=hwfS
-----END PGP SIGNATURE-----

For reference, my libvirt managed virbr0 has forwarding delay 0. This is the default:

http://libvirt.org/formatnetwork.html#elementsConnect

I know that my VM is a leaf node (it only has one NIC and isn't going to create a loop in the network) and therefore it makes sense to eliminate the forwarding delay entirely.

You might be interested in this link about STP delay on physical networks:

http://www.cisco.com/en/US/products/hw/switches/ps700/products_tech_note09186a00800b1500.shtml

Stefan