Bootp woes. 911!

Hello,

We are having multiple network booting problems. We have a server with the
June 2000 version of QNX on it. Using standard Bootp protocol with 3Com
cards in the server and Ziatech cards as blind nodes with Bootp ROMS from
LANWorks. We have some blind nodes (8) connected to a 10/100 switch that is
uplinked to another 10/100 switch with more nodes (8) and then uplinked to a
third 10/100 switch where our boot server is located.

We cannot get several of the blind nodes to finish booting. Here is the
output from “sin” on the boot server.

SID PID PROGRAM PRI STATE BLK CODE DATA
– – Microkernel — ----- — 10448 0
0 1 /boot/sys/Proc32 30f READY — 118k 749k
0 2 /boot/sys/Slib32 10r RECV 0 53k 4096
0 4 /bin/Fsys 10r RECV 0 77k 27115k
0 5 /bin/Fsys.eide 22r RECV 0 61k 258k
0 8 idle 0r READY — 0 40k
0 17 file://1/bin/Dev32 24f RECV 0 32k 98k
0 23 file://1/bin/Pipe 10r RECV 0 16k 65k
0 24 file://1/bin/Dev32.ansi 20r RECV 0 40k 94k
0 28 file://1/bin/Dev32.pty 20r RECV 0 12k 57k
0 31 file://1/bin/Dev32.ser 20r RECV 0 16k 24k
0 32 file://1/bin/Net 23r RECV 0 32k 143k
0 34 file://1/bin/Net.ether905 20r RECV 0 45k 86k
0 35 file://1/bin/Net.ether905 20r RECV 0 45k 90k
0 37 file://1/bin/nameloc 20o RECV 0 6144 20k
0 38 file://1/bin/nameloc 20o REPLY 0 6144 20k
0 328 file://1//usr/ucb/bootpd 10o RECV 2892 40k 49k
0 769 file://1/bin/buildqnx 10o REPLY 23 5120 159k
0 1277 file://1/
/usr/ucb/tftpd 10o REPLY 2021 8192 32k
0 1696 file://1//tcpip/bin/tst 15f REPLY 0 40k 32k
0 1994 file://1/bin/Fsys.floppy 10o RECV 0 20k 40k
0 1997 file://1/bin/Dosfsys 10o RECV 0 49k 73k
0 2021 file://1/
/usr/ucb/Socklet 10r RECV 0 114k 172k
0 2091 file://1/usr/pbin/xqts 4r RECV 0 24k 20k
0 2092 file://1//usr/ucb/inetd 10o RECV 2095 36k 45k
0 2121 file://1/bin/cron 10o RECV 0 24k 20k
0 2122 file://1/bin/tinit 10o WAIT -1 16k 28k
0 2213 file://1/
/tcpip/bin/qst 15f REPLY 0 36k 32k
0 2219 file://1//tcpip/bin/qct 15f RECV 0 36k 32k
0 2222 file://1/
/tcpip/bin/tqt 15f RECV 0 24k 26578k
1 2302 file://1/bin/ksh 10o REPLY 17 15k 45k
0 3043 file://1//usr/ucb/tftpd 10o REPLY 2021 8192 32k
0 3046 file://1/bin/buildqnx 10o REPLY 23 5120 159k
0 4190 file://1/
/usr/ucb/telnetd 10o RECV 14945 53k 53k
0 5224 file://1//usr/ucb/tftpd 10o REPLY 2021 8192 32k
0 15675 file://1/
/usr/ucb/tftpd 10o REPLY 2021 8192 32k
0 19049 file://1/bin/ksh 10o WAIT -1 15k 28k
0 20736 file://1/bin/ksh 10o WAIT -1 15k 28k
0 21098 file://1/bin/buildqnx 10o REPLY 23 5120 159k
1 21994 file://1/*/usr/bin/syslogd 10o RECV 0 36k 32k
0 27964 file://1/bin/ksh 10o WAIT -1 15k 28k
0 27965 file://1/bin/buildqnx 10o REPLY 23 5120 159k
2 28770 file://1/bin/ksh 10o WAIT -1 15k 45k
2 32145 file://1/bin/sin 10o REPLY 1 45k 49k
0 32228 file://1/bin/ksh 10o WAIT -1 15k 28k


Why is buildqnx blocked on pipe? why is tftpd blocked on Socklet?


Kevin Hammond
PACE Control Systems.

“Kevin Hammond” <kevinh@paceolm.com> wrote in
news:a3mbvo$sa5$1@inn.qnx.com:

Hello,

We are having multiple network booting problems. We have a server with
the June 2000 version of QNX on it. Using standard Bootp protocol with
3Com cards in the server and Ziatech cards as blind nodes with Bootp
ROMS from LANWorks. We have some blind nodes (8) connected to a 10/100
switch that is uplinked to another 10/100 switch with more nodes (8)
and then uplinked to a third 10/100 switch where our boot server is
located.

Have you tried moving the nodes into closer proximity (network wise)? The
latency introduced by all those switches maybe enough to make tftp time out
(as tftp is pretty finicky).


\

Cheers,
Adam

QNX Software Systems Ltd.
[ amallory@qnx.com ]

With a PC, I always felt limited by the software available.
On Unix, I am limited only by my knowledge.
–Peter J. Schoenster <pschon@baste.magibox.net>

Why is buildqnx blocked on pipe? why is tftpd blocked on Socklet?



Adam Mallory <amallory@qnx.com> wrote in message
news:Xns91AC5E4EC36D8amalloryqnxcom@209.226.137.4

“Kevin Hammond” <> kevinh@paceolm.com> > wrote in
news:a3mbvo$sa5$> 1@inn.qnx.com> :

Hello,

We are having multiple network booting problems. We have a server with
the June 2000 version of QNX on it. Using standard Bootp protocol with
3Com cards in the server and Ziatech cards as blind nodes with Bootp
ROMS from LANWorks. We have some blind nodes (8) connected to a 10/100
switch that is uplinked to another 10/100 switch with more nodes (8)
and then uplinked to a third 10/100 switch where our boot server is
located.

Have you tried moving the nodes into closer proximity (network wise)? The
latency introduced by all those switches maybe enough to make tftp time
out
(as tftp is pretty finicky).


\

Cheers,
Adam

QNX Software Systems Ltd.
[ > amallory@qnx.com > ]

With a PC, I always felt limited by the software available.
On Unix, I am limited only by my knowledge.
–Peter J. Schoenster <> pschon@baste.magibox.net

Kevin Hammond <kevinh@paceolm.com> wrote:

Why is buildqnx blocked on pipe? why is tftpd blocked on Socklet?

buildqnx writes its output to a pipe, which is read by tftpd.
If the pipe is full, the write() to the pipe will block.

tftpd would normally be blocked on socket – it is waiting for
a network transaction to complete, or another request to come in.

Both of those are relatively “normal” states for the pieces to be
in.

-David

QNX Training Services
http://www.qnx.com/support/training/
Please followup in this newsgroup if you have further questions.

We use this Bootp configuration at several other locations and have not seen
this sort of problem before.
Sometimes the boot request is received and processed immediately and all is
well, but, more often than not,
the tftp image transfer hangs right in the middle and there’s no boot.

We’ve tried various network configurations: boot server and client on the
same switch with short cables
(works ok most of the time, but still hangs 2 out of 10 times), client
connected to a switch that is uplinked
to the switch local to the boot server (works less). And if i try to boot 2
pcs at the same time, no go on either.

“Kevin Hammond” <kevinh@paceolm.com> wrote in
news:a3rgqi$n1j$1@inn.qnx.com:

We use this Bootp configuration at several other locations and have not
seen this sort of problem before.
Sometimes the boot request is received and processed immediately and
all is well, but, more often than not,
the tftp image transfer hangs right in the middle and there’s no boot.

We’ve tried various network configurations: boot server and client on
the same switch with short cables
(works ok most of the time, but still hangs 2 out of 10 times), client
connected to a switch that is uplinked
to the switch local to the boot server (works less). And if i try to
boot 2 pcs at the same time, no go on either.

Was this test conducted after the switches where allowed to normalize? Did
the similar setups in other locations use the same switches? The same
NICs?


\

Cheers,
Adam

QNX Software Systems Ltd.
[ amallory@qnx.com ]

With a PC, I always felt limited by the software available.
On Unix, I am limited only by my knowledge.
–Peter J. Schoenster <pschon@baste.magibox.net>