Server Download File Slowdown Question

[Mike Luther]

I have an MCP2 latest everything SCSI 160 Level hard drive with an Adaptec
2940U2W Pro and Seagate 336754LW hard drive plus a KIngston 120 NIC card,
which unfortunately, for the while now, has been and will have to work with
both 10 and 100 level network connectivity as a server with PEERLAN here.
This particular older unit has been working very reliably for years now with
Kingston network cards or Intel Mobo NICs with the exception of the one very
old Coax/10 level NIC that is still on one system connected to this network.
Strangely, uploads from the 10 Level old box to the server work flawlessly at
the expected 10-Level LAN rate. As well as uploads to the server from any of
the other modern units - but at a 10 Level speed! However, when I go to
download anything from this server to ANY other box, including the old
10-Level unit, all I ever see for larger files, especially 100MB files which I
have to handle even several times a day between these units ..

It can take a HORRIBLE long time to even download the file(s)!

All I see is flicker -- time -- flicker -- time -- flicker on the switch and
router for the LAN until the whole file is finished. No problem with
integrity - just HORRIBLE slow LAN rate. Plus the CPU load is so low that it
never even gets off the bottom line for any of this transfer. So that is, I
think not a problem here.

I've gone and deliberately used the GammaTech HPFS OS/2 Defrag utility to, for
example, completely optimize the hard drives on the server and the workstation
before a download to a workstation. No help.

Can anyone here suggest what else may be causing this issue? I don't see this
same issue on any of several other sites I orchestrate which have the same
general type configuration. But not with any old COAX/10K level NIC use.

Can anyone here suggest how I might research and fix this issue?

Thanks!

[j***@nospam.com.au]

Post by Mike Luther
I have an MCP2 latest everything SCSI 160 Level hard drive with an Adaptec
2940U2W Pro and Seagate 336754LW hard drive plus a KIngston 120 NIC card,
which unfortunately, for the while now, has been and will have to work with
both 10 and 100 level network connectivity as a server with PEERLAN here.
This particular older unit has been working very reliably for years now with
Kingston network cards or Intel Mobo NICs with the exception of the one very
old Coax/10 level NIC that is still on one system connected to this network.
Strangely, uploads from the 10 Level old box to the server work flawlessly at
the expected 10-Level LAN rate. As well as uploads to the server from any of
the other modern units - but at a 10 Level speed! However, when I go to
download anything from this server to ANY other box, including the old
10-Level unit, all I ever see for larger files, especially 100MB files which I
have to handle even several times a day between these units ..
It can take a HORRIBLE long time to even download the file(s)!
All I see is flicker -- time -- flicker -- time -- flicker on the switch and
router for the LAN until the whole file is finished. No problem with
integrity - just HORRIBLE slow LAN rate. Plus the CPU load is so low that it
never even gets off the bottom line for any of this transfer. So that is, I
think not a problem here.
I've gone and deliberately used the GammaTech HPFS OS/2 Defrag utility to, for
example, completely optimize the hard drives on the server and the workstation
before a download to a workstation. No help.
Can anyone here suggest what else may be causing this issue? I don't see this
same issue on any of several other sites I orchestrate which have the same
general type configuration. But not with any old COAX/10K level NIC use.
Can anyone here suggest how I might research and fix this issue?

I would compare the MCP and LAN configs between good and bad boxes.

It is so long since I have looked at mine that I cannot recall the names of the
relevant files.

[Mike Luther]

Thanks John!

Post by j***@nospam.com.au
I would compare the MCP and LAN configs between good and bad boxes.
It is so long since I have looked at mine that I cannot recall the names of the
relevant files.

Problem found and solved! PCI based NIC card was strangely defective in this
box. What was happening was the interface CHIP between the CAT-5 socket and
the card had gone defective and was hitting the card back internally with
strange pulses that stopped the outbound data traffic on that wire in the
CAT-% cable until the twittering stopped. The card would service outbound
data traffic again until this twitter started. Then on and on back and forth.
I do have and have used COMM port tool research to trace every byte and so
on of POTS phone line data over years here with BBS systems and so on. But
never tried to do that with a LAN connection to trace byte by byte what was
going wrong one direction, but not another.

Amazing...