froelich@ux1.cso.uiuc.edu (Bill froelich) (04/09/91)
He at the Univ. of Illinois in one of our student labs, I noticed an interesting statistic in fconsole the other day and was wondering if anyone could clue me in to it's meaning. Setup: Netware 2.15c on a gateway 2000 server Approximately 95 active users - split Mac and PC. All but 20 macs are on ethernet and the others are on localtalk. All the PC's are on ethernet to the server. Server is running Mac Vaps 1.1 Question: I was looking at the fconsole statistics summary page and there is the current utilization field. This number was consistently around 170-180% It would drop to around 30-40% for a few seconds and then right back up to where it was. It stays around 175% when the lab is open, (most of the day) It has been as high as 190%. What does this number tell me? Do I need another server to handle the load? The reason I looked was that the macs would "freeze" up every now and then. The watch would come on and nothing would happen for a few minutes, then everything would be fine again for a while. It would then freeze again. I thought maybe someone was accessing the server for a huge file but the utilization is still 170-180% even now with everything apparently working fine. Any suggestions? I apologize if this has been talked about before. ------------------------------------------------------------------------------ Bill Froelich Univ. of Illinois Site Manager froelich@ux1.cso.uiuc.edu froelich@uiucvmd.bitnet
sylvia@brahms.udel.edu (Sylvia M Berta) (04/12/91)
I have no answer to the problem, but do have a related (and occasionally fatal) dilemma. We have a lab with about 30 some IBM 30's and Zenith 386sx's, plus 16 Mac's. The server is an IBM PS/2 model 80 with 4M RAM and 2 hard drives (a 70M and a 310M, so not mirrored), using UTP and Cabletron cards and packet drivers (for TCP, etc.). The problem is: recently (and intermittently) the system has been crashing with a General Protection Error the only message that's has appeared. We've replaced several memory chips, and the mother board. We've also reinstalled the operating system, which is 2.15c with the Mac VAP. We did discover that the Current Server Utilization shot up to over 90% prior to one of the crashes and stayed there for several minutes (an assistant was looking at FCONSOLE because the system was slow). Normally we run around 20-30% under a full load of users, and that's only peaks. There was also a Mac print job that never finished, after 25 minutes! It seems to be a hardware problem, but I don't seem to be able to track it. Any help or suggestions would be appreciated. On another system, we've just recently noticed that SYSCON is giving us strange output in the USER INFO/name/OTHER INFO option where you should be able to discern the total amount of disk space allocated and the amount used. We see the amount allocated, but the amount used always appears as 4k more the the amount allocated! This is for every user we have...but it doesn't operate that way. That is, no one has gotten any messages saying that they had overextended their limit. And VOLINFO doesn't show any extra space being used. For reference, we are running 2.15c (waiting for 3.11)...again on a model 80 (20 Mhz machine). We lost our hard drive a few weeks ago when the Hot Fix area overloaded and shut off, and the hard drive failed the COMPSURF tests. The drive was replaced and the system reinstalled (I can install 2.15c in my sleep!) 8-) Again, any help or suggestions appreciated. The only other thing I can think of at this point is to run BINDFIX, and then VREPAIR, but things have been so strange lately that it makes the Site Director very nervous if I suggest taking the system down when it appears to be serving us properly (with the exception of the SYSCON glitch). Thanks in advance ... Syl Berta
ileader@sideways.gen.nz (Ian Leader) (04/12/91)
> Question: > I was looking at the fconsole statistics summary page and there is the > current utilization field. This number was consistently around 170-180% > It would drop to around 30-40% for a few seconds and then right back up to > where it was. It stays around 175% when the lab is open, (most of the day) > It has been as high as 190%. > What does this number tell me? Do I need another server to handle the load? > > Bill Froelich Univ. of Illinois > Site Manager > froelich@ux1.cso.uiuc.edu froelich@uiucvmd.bitnet I'd be keen to hear if someone has an answer to this one too. Up until a few weeks ago, our 2.15 server was a 25MHz 386 machine and the utilization percentage showed what I would expect it to: varying between 1 and about 70-80% as the load on the server changed. I then moved the entire system (hard disks and all) into a 33MHz machine of exactly the same brand and imediately after bringing the server up the first time, it showed utilisation figures of around 170% with the system idling and no-one logged in. Occassionally it will flick back to 30-40% when the disk i/o goes up. I have also noticed that the number of open files shows as about 17 even with no workstations logged in, increasing from that number as each file is opened by a workstation. This to me appeared to be some sort of corruption becuase a utilisation over 100% seems like nonsense. No-one so far has been able to shed any light on it here. Can anyone out there help? Certainly, I don't think it is an indication of the need for a more powerful machine!? Ian Leader - Avalon Television Centre - Lower Hutt - New Zealand
saddison@ca.excelan.com (Skip Addison) (04/13/91)
The News Manager) Nntp-Posting-Host: ca Reply-To: saddison@ca.excelan.com (Skip Addison) Organization: Novell, Sunnyvale, CA References: <1991Apr8.191248.26526@ux1.cso.uiuc.edu> <55g9Z1w164w@sideways.gen.nz> Date: Fri, 12 Apr 1991 14:40:26 GMT In article <55g9Z1w164w@sideways.gen.nz> ileader@sideways.gen.nz (Ian Leader) writes: >> ... It stays around 175% when the lab is open, (most of the day) >> It has been as high as 190%. >> What does this number tell me? Do I need another server to handle the load? >> >> Bill Froelich Univ. of Illinois >> Site Manager >> froelich@ux1.cso.uiuc.edu froelich@uiucvmd.bitnet > >I'd be keen to hear if someone has an answer to this one too. Up until a >few weeks ago, our 2.15 server was a 25MHz 386 machine and the >utilization percentage showed what I would expect it to: varying between >... I haven't seen any replies yet, so I'll take it. On "fast" machines some releases of NetWare (including 2.15c, but I forget what others -- maybe all) report erroneous utilization figures. Fast machine ::= 386/25 or higher. It is a cosmetic problem only. It's being fixed, but it's not been a very high priority since it is cosmetic only and in most installations NetWare is very much I/O bound, not CPU bound. In our 250-workstation demonstration at NetWorld last year with all the Mac, DOS and OS/2 workstations, processor utilization never got above 20-25%. I hope this helps. -- Skip saddison@novell.com
klimas@iccgcc.decnet.ab.com (04/17/91)
In article <55g9Z1w164w@sideways.gen.nz>, ileader@sideways.gen.nz (Ian Leader) writes: >> Question: >> I was looking at the fconsole statistics summary page and there is the >> current utilization field. This number was consistently around 170-180% >> It would drop to around 30-40% for a few seconds and then right back up to >> where it was. It stays around 175% when the lab is open, (most of the day) >> It has been as high as 190%. >> What does this number tell me? Do I need another server to handle the load? >> >> Bill Froelich Univ. of Illinois >> Site Manager >> froelich@ux1.cso.uiuc.edu froelich@uiucvmd.bitnet > > I'd be keen to hear if someone has an answer to this one too. Up until a > few weeks ago, our 2.15 server was a 25MHz 386 machine and the > utilization percentage showed what I would expect it to: varying between > 1 and about 70-80% as the load on the server changed. I then moved the > entire system (hard disks and all) into a 33MHz machine of exactly the > same brand and imediately after bringing the server up the first time, it > showed utilisation figures of around 170% with the system idling and > no-one logged in. Occassionally it will flick back to 30-40% when the > disk i/o goes up. I have also noticed that the number of open files shows > as about 17 even with no workstations logged in, increasing from that > number as each file is opened by a workstation. > This to me appeared to be some sort of corruption becuase a utilisation > over 100% seems like nonsense. No-one so far has been able to shed any > light on it here. Can anyone out there help? > > Certainly, I don't think it is an indication of the need for a more > powerful machine!? > > Ian Leader - Avalon Television Centre - Lower Hutt - New Zealand This is a long time, well known bug in NOVELL Advanced Netware 286 when it is executing on a PC with greater than 25Mhz clock speed. The disturbing thing is that everyone picks up on it immediately and starts getting nervous about the integrity of the system. This problem is very visible and very old and yet NOVELL, has done nothing to fix their code!
jlamb@npd.Novell.COM (Jason "Nematode" Lamb) (04/18/91)
The 100+ percent utilization monitor screen you see is nothing more than a timing bug in the utilization routine with the (I believe) 2.15 rev A. version of Advanced NetWare. This doesn't indicate that anything is wrong except the server is going a little too fast for the routine. This is fixed in the latest version of 2.15c (or it's been fixed in the 2.2. I forget which). Jason Lamb
bryans@ux1.cso.uiuc.edu (B. Charles Siegfried) (04/19/91)
saddison@ca.excelan.com (Skip Addison) writes: >In article <55g9Z1w164w@sideways.gen.nz> ileader@sideways.gen.nz (Ian Leader) writes: >>> ... It stays around 175% when the lab is open, (most of the day) >>> It has been as high as 190%. >>> What does this number tell me? Do I need another server to handle the load? >>> >>> Bill Froelich Univ. of Illinois >>> Site Manager >>> froelich@ux1.cso.uiuc.edu froelich@uiucvmd.bitnet ... >I haven't seen any replies yet, so I'll take it. On "fast" machines some >releases of NetWare (including 2.15c, but I forget what others -- maybe all) >report erroneous utilization figures. Fast machine ::= 386/25 or higher. >-- Skip > saddison@novell.com I work at Bill's site, and we only have a Gateway '286 machine running. I certainly wouldn't classify it as a "fast" machine. __ Bryan Siegfried Bioecon (& Novell too) at UIUC
bryans@ux1.cso.uiuc.edu (B. Charles Siegfried) (04/20/91)
bryans@ux1.cso.uiuc.edu (That's me :-) writes: > I work at Bill's site, and we only have a Gateway '286 machine >running. I certainly wouldn't classify it as a "fast" machine. Oops, I do believe we have a '386. I guess it's too early in the morning. :-) __ Bryan Siegfried Biology, Economics, and Novell at UIUC
kenh@techbook.com (Ken Haynes) (04/20/91)
The server utilization code *WAS* fixed in NW version 2.2. UPGRADE, UPGRADE
--
******************************************************************************
* Ken Haynes, CNE | 1-900-PRO-HELP
* Technical Support Product Manager, 900 Support
* UUCP: {nosun, sequent, tessi} kenh@techbook