Topic: Problems accessing from XP after upgrade 5.1.2 to 5.6u4

[Michael Smith]

Assistance would be appreciated, please!

SME 5.1.2 box holds a LOT of pictures of doctor's patients.  He was having an odd problem with XP's "filmstrip" view; he couldn't rotate any of the pictures with the built-in tool in Windows Explorer; XP thought the files were "read-only" even though I checked the permissions from a shell prompt.  Well, this machine is behind a firewall and server-only so I hadn't worried about upgrading to 5.6, but thought perhaps an upgrade would fix this odd issue.  

Upgrade went fine.  Now XP's behavior is worse!  It takes FOREVER to read the folder (100 megabits all 'round, network-wise).  The original problem is solved ... but it's really unusably slow.

The picture files are organized in subfolders by patient name; there are LOTS of patients but only a few pictures per folder, so I can't see it being a problem with too many files in each folder.

Technical notes:  the server is an HP TC2100.  SME 5.1.2 would not recogize the built-in ethernet adapter so I finessed the issue by putting in a NIC I knew was supported.  5.6 recogized the built-in adapter and assigned it the eepro100 driver.  I note on the support page:

http://h20004.www2.hp.com/soar_rnotes/bsdmatrix/matrix71126en_US.html

that HP offers RH 7.1 drivers for the chipset and the adapter.    Should I install these?  How would I tell SME to use these drivers?

The server is explicitly defined in the HOSTS & LMHOSTS files of the XP machine.

Here's what the Samba log associated with the offending XP box says:

2003/08/25 02:22:55, 0] lib/util_sock.c:read_data(436)
  read_data: read failure for 4. Error = Connection reset by peer
[2003/08/25 02:27:22, 0] lib/util_sock.c:read_data(436)
  read_data: read failure for 4. Error = Connection reset by peer
[2003/08/25 02:29:00, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 02:29:00, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 5: ERRNO = Connection reset by peer
[2003/08/25 02:29:00, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)
[2003/08/25 02:32:10, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 02:32:10, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 12: ERRNO = Connection reset by peer
[2003/08/25 02:32:10, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)
[2003/08/25 02:39:11, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 02:39:11, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 12: ERRNO = Connection reset by peer
[2003/08/25 02:39:11, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)
[2003/08/25 02:52:25, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 02:52:25, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 12: ERRNO = Connection reset by peer
[2003/08/25 02:52:25, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)
[2003/08/25 03:05:31, 0] lib/util_sock.c:read_data(436)
  read_data: read failure for 4. Error = Connection reset by peer
[2003/08/25 03:09:21, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 03:09:21, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 5: ERRNO = Connection reset by peer
[2003/08/25 03:09:21, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)
[2003/08/25 03:14:26, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 03:14:26, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 12: ERRNO = Connection reset by peer
[2003/08/25 03:14:26, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)
[2003/08/25 03:17:10, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 03:17:10, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 5: ERRNO = Connection reset by peer
[2003/08/25 03:17:10, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)
[2003/08/25 03:20:26, 0] lib/util_sock.c:write_socket_data(499)
  write_socket_data: write failure. Error = Connection reset by peer
[2003/08/25 03:20:26, 0] lib/util_sock.c:write_socket(523)
  write_socket: Error writing 61503 bytes to socket 5: ERRNO = Connection reset by peer
[2003/08/25 03:20:26, 0] lib/util_sock.c:send_smb(703)
  Error writing 61503 bytes to client. -1. (Connection reset by peer)

[ryan]

I crashed a small XP domain going from a 5.1.2 pdc to 5.6 pdc and upgrading all XP systems using windowsupdate.  All would have been well had I first installed the registry patch which is located in /home/e-smith/server-resources/regedit/winxplogon.reg on 5.6.   I discovered that XP service pack 1a caused communication problems with the 5.6 pdc.

I built a test XP box.  I applied the reg patch after installing XP, then installed updates after the patch.   This box experienced no problems with domain communications.  Hope that helps your samba issues.

ryan

[Michael Smith]

The SME box is not now and has not been a domain controller.  Simply a server.  The client machine is XP Home Edition.

[Michael Smith]

winxplogon.reg did not help.  I'm strongly suspecting that the eepro100 driver is not wholly appropriate for this chipset.  Any thoughts on how to get SME to recognize the HP-provided drivers, as they are RH 7.1-specific?

[ryan]

Honestly, with price of nics being very affordable, I would not bother messing with drivers on my systems...just put in replacement nic cards.  I am not real savy with linux/bash programming and syntax, so I always go for the simplest fix which does not involve heavy modification of system files or drivers.

ryan

[Kevin Tollison]

I am running Dell 600SC with eepro1000 (Gagabit).  These drivers took care of my issues.  The 100 and 1000 drivers are available.

http://mirror.contribs.org/smeserver/contribs/dmay/mitel/contrib/Intel/sme56/

[Michael Smith]

Might have to chalk this one up to an ID10T error, or IFS ("Insufficient Forum Search").  I found my old thread here:

http://e-smith.org/bboard/read.php?f=3&i=14007&t=13986&v=f

from when I first set up this server!!!!  Oh well, I was hoping that later drivers would handle it.  I did download & install the e100 RPM, also the procfg & procfgd.  e100 installed fine & I set it in the configuration screens booting as "admin".  Still slow as molasses.  I tried to install procfg & procfgd, they demanded libcrypto, I installed w/nodeps, no dice.

Ryan, you're right, why fight it?  It irks me not to have the actual issue resolved, but the important thing is that the doctor not get impatient with his computers.

One reinstall of the SMC 1211TX card & reconfiguration later, all's well.  At least until two years from now when I upgrade to 7.0 and go through the whole rigmarole again.

Thanks for the replies, y'all.

[Ed Form]

Michael Smith said...

> Assistance would be appreciated, please!

> SME 5.1.2 box holds a LOT of pictures of doctor's patients. He was having an
> odd problem with XP's "filmstrip" view; he couldn't rotate any of the pictures
> with the built-in tool in Windows Explorer; XP thought the files were "read-only"
> even though I checked the permissions from a shell prompt. Well, this machine
> is behind a firewall and server-only so I hadn't worried about upgrading to 5.6,
> but thought perhaps an upgrade would fix this odd issue.
>
> Upgrade went fine. Now XP's behavior is worse! It takes FOREVER to read
> the folder (100 megabits all 'round, network-wise). The original problem is
> solved ... but it's really unusably slow.

> The picture files are organized in subfolders by patient name; there are LOTS
> of patients but only a few pictures per folder, so I can't see it being a problem
> with too many files in each folder.

While I wouldn't rule out NIC trouble, I have had very similar problems from 2 aspects of SME when image viewing. My main business is providing document imaging services to clients, and have used SME as my only server since 4.1.2, so I've had quite a bit of experience in this area.

1. Windows XP home is completely unsuitable for this kind of work, and you should bin it.

One problem I found with XP Home was mis-reporting of the numbers of images in directories on SME servers of all versions - literally it just doesn't see all of the files, often up to 20% are apparently missing. Yet copy the archive to a windows machine, and look at it with the same XP home workstation, and it is complete. A Windows XP Pro workstation always reports the correct number of files.

2. The second problem that gives trouble when image handling from SME 5.6 servers, is extreme, and variable, slowness in moving data into a processing program running on a Windows machine. Things like batch conversion from G4-tiff to PaperPort .max files tend to run like treacle. In earlier SME versions such conversions would start slowly but pick up speed after a short time, and run well for the rest of the session. In 5.6 the slowness persists throughought the batch. This kind of problem applies to both XP Home and Pro workstations. SME 5.1.2 did not have the problem and the 6 betas seem OK also. The problem is definitely to do with network transport problems, but it seems to be something to do with caching. When a bit of the file is being brought off the server, then processed, and the next bit of the file requested, the process looses its way and long delays occur. Eventually the system seems to learn where everything is and the processing rate jumps by a factor of 10 or more.

I reported this problem as a bug but was told that the Mitel folks couldn't reproduce it - where they would have obtained a really big image archive from to test it I can't imagine!!!. My suggestion is that you might be seeing something similar.

See http://forums.contribs.org/index.php?topic=17089.msg66298#msg66298 and http://forums.contribs.org/index.php?topic=17089.msg66298#msg66298 for details.

Finally, the filmstrip viewer in Windows XP is a joke. Buy your client a copy of something like PaperPort.

Ed Form

[Michael Smith]

Thanks for your informed response, Mr. Form!  You're right, XP Home is an even smellier pile of you-know-what than XP Pro, but the doctor didn't want to spend the extra $$$.  

All's well again with another NIC in the box.  There are only a few files per folder but there are a LOT of folders, so the issue of not seeing all the files in a given folder doesn't arise.

The doctor likes the filmstrip viewer.  Who am I to say him nay?  He's happy, especially now that he can rotate the images at will.  (That was the problem that precipitated the whole thing!)

[Ed Form]

Michael Smith wrote:
>
> Thanks for your informed response, Mr. Form!  You're right,
> XP Home is an even smellier pile of you-know-what than XP
> Pro, but the doctor didn't want to spend the extra $$$.

Snip

Always good when something has a simple solution.

Ed Form