XP client 4.83 Netware 5.0 slow file open / copy

make sure that all client-side caching is turned off.  There is a setting for FILE CACHING in the cleint software settings.  There is also a server-side command ("SET CLIENT FILE CACHING ENABLED=OFF", IIRC)

Thanks for the suggestions.  BudDurland, I turned off client side cacheing, but it had no effect.  What does the IIRC in your comment refer to?  DSPoole, I've gone through the suggestions on the page you linked to.  I had been there before, but I looked a bit more closely this time.  No help yet.  
I do have a bit more information to report though.  I tried copying my test file (1.3 megs) from the local workstation to the novell drive.  It took about 5 seconds.  Still slow, but not the 30 seconds it takes to copy from the network to the local drive.  I also tried copying a 1.7 meg file from a drive mapped to a NT server sitting on another LAN segment.  It's a pure IP mapping that goes through a router.  That copy took about 1 second.  That makes me think that I need to get rid of IPX.  I'm going to reinstall the client IP only and see what happens.

Steve

I'm seeing this same behavior on my 2000 clients.  Curiously, my personal 2000 machine was working OK yesterday, but today it was slow just like the XP machine.  The only change I made was the SET CLIENT FILE CACHING ENABLED=OFF.  But I had already set it back on before I did the test.  There may have been some MS critical updates, though.

It looks like a server problem now.  I mapped a drive to our Border Manager machine, and the access speed is instantaneous.  The server needs a reboot just now.  I'll be doing that after 5 PM.

stevepep,

Why are you using IPX? NetWare v5 doesn't need it (it can support it, but it doesn't need it).

server needs a reboot????

what are your server stats? (memory and cache buffers, etc.)

you may not have enough RAM.

I wonder if some OS files have become corrupted.  Last Sunday AM in the midst of Mcafee netshield all file scan, the a process aborted and Monday morning the server was inaccessable until it was rebooted.  Prior to that, there were several times that saving files from Word resulted in them being locked with processor utilization hovering at 65%.  Today, it has taken me 15 to 20 minutes of waiting before NWAdmin would start. The server is currently running at 35% utilization.  It typically settles down around 5% with periodic spikes. As I think about it, this file transfer slowness has been dogging us for many months, and I had gotten temporary relief at my W2000 workstation by upgrading to the latest Novell client.  However, Windows critical updates seemed to degrade performance.  Today I mapped a drive to my border manager server and copied a 2 meg file (instantaneously), and I concluded that there are some server issues here.  Here are some server stats:

Original Cache Buffers:   114,064
Total Cache Buffers:       75,615
Dirty Cache Buffers:            5
Current Disk Requests:          0
Packet receive buffers:       128
Directory Cache Buffers:      292
Current Service processes:     38
Current connections:           64
Allocated Memory pool: 84,828,160 - 18%
Cache Buffer memory:  310,210,560 - 66%
Cache movable memory:            0
Cache non-movable memory:   90,112
Total Server work memory: 467,193,856  

Yes, I know that it is unusual to have to reboot Novell servers.  Right now, I'm wondering if re-applying support pack 6a would be helpful.

what are your current Cache Utilization Statistics?

McAfee eh?  Might as well just load the server up with viruses ;)

Short term cache hits:  100%
Short term cache dirty hits: 100%
Long term cache hits     94%
Long term cache dirty hits: 99%
LRU Sitting time 2:02:20.2
Allocate block count: 5,141,684
Allocated from AVAIL:   691,25,
Allocated from LRU:  4,451,656
Allocate wait:  0
Allocate Still waiting: 0
Too many dirty blocks: 0
Cache reCheckBlock count: 0

well - it's definitely not a RAM issue.

I would reapply the service packs and any post-SP patches and then reboot to see if that helps.  You may have a corrupted file ala' McAfee.  If you can, maybe you ought to configure it so it ignores the SYS:\SYSTEM folder?

Short term cache hits:  100%
Short term cache dirty hits: 100%
Long term cache hits     94%
Long term cache dirty hits: 99%
LRU Sitting time 2:02:20.2
Allocate block count: 5,141,684
Allocated from AVAIL:   691,25,
Allocated from LRU:  4,451,656
Allocate wait:  0
Allocate Still waiting: 0
Too many dirty blocks: 0
Cache reCheckBlock count: 0

Sorry about the double post.  Aparently the site had a glitch since it told me to retry.

Looks like I'll be staying late to do some patching.  Thanks for the Mcafee comments.  I'll check the configuration.

Did you do the NW client install with Custom selected and IP only??  Do you have SLP instaled on the server and setup on the clients??

Thanks for the comment d50041.  Yes, I did the custom install.  I didn't look for SLP on the server since I was able to log in.  I'm now convinced this is a server issue, and I'll be working on the server starting at 5 pm CST.

I've had little problems with Computer Associates InoculateIT 4.5 for NetWare...

that and the sig file updates are FREE for life.

Try this one,
http://support.novell.com/cgi-bin/search/searchtid.cgi?/10068696.htm

Thanks, duncs for the reference.  I'll try it tomorrow.  I just finished dsrepair, vrepair, reinstall SP6a.  It didn't help.  I still think it's a server problem since I can open files on other servers instantly.  Also, it's a one way problem.  I can copy files to the servers quite rapidly.  The expanded SP6a is over 400 megs, and I copied it to the server in a couple of minutes.  It takes 10 minutes to open NWAdmin, though.  I'm hearing from a lot of users that their apps are slow.

Another approach is to down the server and bring it up "server -na", then bring up additional ncf files you hace one at a time and try to determine if any one particular program is causing the utilization issues. Especially programs such as your virus scan....IMHO I do not run any virus scan on my servers, just on the desk tops.  Not right or wrong, better or worse just less burden on the server and no one accesses it directly except me so anything on it has to go through a user workstation anyway.

Thanks for the suggestion d50041.  I'm going to try that.  Nothing I like more than staying after work to troubleshoot a server. ;-)  Last night after re-installing SP6a I brought up the server, then unloaded anti virus and groupwise.  I then booted a couple of workstations to test and found no change in behavior.  Here's a summary of observed symptoms so far:

1) Using client 4.83 on XP or 2000 copy 1.3 meg file from server to workstation takes 30 seconds.  A windows dialogue box appears indicating that 45 seconds remain, then the graphic progress indicator fills slowly and the whole process takes about 30 seconds.

2) browsing is quite snappy.  Many of the speed tips suggested here are for delays in browsing.  I am not experiencing that.

3) Copying files to the server works fine.  To install SP6a, I first tried expanding the compressed distribution package on the server, but it became apparent that the delay in getting data from the server to the workstation would have left me waiting for hours as it verified the package.  I expanded the file on my workstatation and copied over 400 megs to the server in under 2 minutes.

4) I can map a drive to my border manager server and copy files without difficulty.  A 2 meg text file copies instantly.  No dialogue box.  Just drag and drop, and it's there.

5) I had been noticing delays in opening excel and word files for months.  After troubleshooting that problem on a W98 machine and fixing it by (per a novell tid) installing the latest client, the problem recurred when I installed Windows updates.  I noticed that file open speeds seemed to change with each windows update.  Most recently, my W2000 workstation was ok last week, but on Monday of this week had the same problem as the XP box I am trying to get ready to deploy.

6) I have had about a half a dozen instances of files getting locked when saving.  The workstation will after a delay appear to have successfully saved the file to the network, but trying to access the file will be unsuccessful, and the workstation will hang untill everything times out.  Viewing file open / lock activity in monitor will show files in use.  Even after shutting down the workstation and re-booting, monitor will show the former connection as active, and monitor will be unable to delete the connection.  Processor utilization on the server will spike to 65% or so, and not settle down till the server is downed and restarted.  Last night, however, there were files locked in this manner, but utilization was "only" at 35%.

I don't think that I'll try "server -na" tonight since the power company is re-doing our entrance and cutting our power at 4:00 AM.  I'll need to be in early to restart everything, so this problem will have to wait a bit.

I like that idea.  The problem has been getting steadily worse, and that's typical of hardware problems.  Users have been complaining of sluggish app opening, and I fixed some by moving them from a hub to a switch.  They got much faster, but within a couple of weeks, they were slow again.

I have a gigabit NIC for the server that I was planning on installing as soon as the wiring contractor pulls some cat 6 cable for me.  I think I should put in a new 100 mb card right away.  I don't see anything that could be causing electro-mechanical interference, but I'll check that out also.  The fact that the Border Manager server is right next to the other server, and has been for many months argues against EM interference as the culprit.

I suppose that the relatively better performance on windows 98 workstations (can copy my test file in a couple of seconds) could be accounted for by different protocol implementation.

might as well try swapping that NIC out to see if that helps - could also be the XP hardware on that switch as well.

I would wonder what would happen if Windows 2000 were loaded on those XP machines - are they better or worse?  If better, then the OS is crap.  If worse/same - then hardware on the computer is crap.

I notice the Compaqs have an Intel PRO/100 VM NIC on board, the V-series is capable of low-power usage - is it possible to reconfigure the hardware not to use low-power specs?

The XP machines are Compaq Evo D510s.  I also have some Evo D500s and I believe (both have Intel on embedded cards)  They are currently all experiencing the same symptoms.  What is confusing is why my 2000 machine was performing better initially than the XP machine (now the same).  I think it's worthwhile to try a new NIC.  If I can get one today, I'll try putting it in tonight.  Maybe the gigabit card can be configured to run at 100baseT.

I'll check the bios settings on the onboard card, but if I'm downing the server, I may as well just put in another NIC.

Interesting that the problem seems to worsen over time, have you looked at NIC stats on the server for anything out of the ordinary??

Total packets transmitted    4,906,304
Total packets received       5,260,717
Recieve failed adapter overflow condition  85
Transmit failed misc. error                 2

Since 8 pm last night.  

Those don't seem like damning stats to me.

I put anoter NIC into the server last night and it solved the problem.  Thanks, DSPoole for first suggesting it.  I'll be awarding you the points.

The NIC install was an interesting adventure.  I had purchased a 3Com 3C996B gigabit card and told our wiring contractor that he could pull some cat 6 when he got around to it.  I tried to install it to run at 100mb.   First, the server didn't recognize the driver disk made from the install CD.  Then, after manually copying the drivers to the Sys: volume and binding TCP/IP, the server locked up when it tried to load the drivers.  After disabling the NIC in InetCfg I brought up the server, logged in from a workstation, gave up for the evening thinking I'd try a different card the next evening.  Later that evening, we powered down the server for a couple of hours to accomadate utility work on our building, the server came up the next morning with monochrome video.  This was a problem for backup exec because the colors used for fonts and background all rendered as the same shade of grey.  Curiously, RConsole displayed the backup exec screens in color, but other screens in monochrome.  

After removing the 3Com card on the theory that perhaps it added something to the BIOS that conflicted with the video settings and re-booting, the server was fine.  Since I didn't have a server card, I put in a 3C905C workstation card, and configured it.  It couldn't have gone more smoothly.  Now, all that remains is to get a server card (Compaq has a gigabit card that is supposed to be compatible with the Proliant 1600).  And - oh yes - the XP box doesn't browse as quickly as the 2000 boxes.  I think I'll have to start at the top of this thread again.  

Thanks to all who contributed.  

Steve

Thanks for the input.  I wouldn't have pulled all the symptoms together in a list as readily without this forum.

Steve

Steve,
I have exact the same problem as you have. Did you solve this problem? What did you do? I have already implement the last service pack of the client (updated last week). Please let me know, because browsing and copying is also very slow to our novell server.
Thanks,
Ray