{{tag>category:"Network Product" author:"dsig@sigafoos.org" author:"Gray Cunningham" author:" [url=http://www.sprezzatura.com]The Sprezzatura Group[/url]" author:"d sigafoos" author:"Richard Bright" author:"Don Miller - C3 Inc." author:"John Bouley" author:"Dave Harmacek" author:"Pat McNerthney"}} [[https://www.revelation.com/the-works|Join The Works program to have access to the most current content, and to be able to ask questions and get answers from Revelation staff and the Revelation community]] ==== arev / fileservice slowdown cripples system (Network Product) ==== === At 12 FEB 2003 11:44:40AM dsig@sigafoos.org wrote: === NOTE: i have posted this on the Open site also I got a call from a guy whose system periodically slows down to a crawl. This can happen when a lot of users are on or when just a couple (2 or 3) are running. It can even happen when a single station is running the nightly processes. Happy to get any additional information needed .. Arev 3.12 with the 1.5 Windows fileservice on an NT4.0 server. Workstations are all win98 (i believe). PIII 950 with 512meg memory. Disks are stripe 3 and they have 30-40 gig of free space. Arev service is the only thing running on the server. When this happens the Disk Queue shoots way up .. the processor is never chanllenged. They had thought they had disk problems some months ago but those have been replaced. the problem still occurs. I have searched the archives and have not found anything relevant here except a bug back in 96 dsig@sigafoos.org onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)" David Tod Sigafoos ~ SigSolutions Phone: 971-570-2005 OS: Win2k sp2 (5.00.2195) OI: 4.1.2 ---- === At 12 FEB 2003 01:02PM Gray Cunningham wrote: === Dsig, I am very interested in hearing about a solution to this problem as one of my customers is experiencing the same problem. However, they are running a Win 2000 Server with v2.1 of the 2000 Service and XP Workstations, and Arev 3.12. Their network guy has been in contact with MicroSoft and they had him send them some sort of huge diagnostic file that they are allegedly looking at...but it has been over a week and still no answer. If I do hear anything, I'll post it here and I would appreciate it if you would post anything that you find out. Good luck, Gray Cunningham ---- === At 12 FEB 2003 01:22PM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote: === Gray We'll ask the same of you as we have of Sig privately. Largest file size? LK/OV/Modulo/Row Count? [url=http://www.sprezzatura.com" onMouseOver=window.status= Click here to visit our web site?';return(true)]The Sprezzatura Group[/url] [i]World Leaders in all things RevSoft[/i] [img]http://www.sprezzatura.com/zz.gif[/img] ---- === At 12 FEB 2003 02:14PM d sigafoos wrote: === Below is a some background info from the user along with some system/network information and workstation information. Looks pretty 'plain jane' to me He does have a couple of files that are way overflowed. The biggest bing 700K in ov and 100K in lk. Sizelock not set .. I am writing a quick little 'stats' routine for him to get the record cnts/sizing etc and checking the keys at the same time. DSig Here?s everything I can think of that may be relevant. Please let me know what else you would like to know. Symptoms: System used to have periodic episodes of slow performance, that would last 30 minutes or so, where everyone using the system would experience very slow response time. We could never pin down what was causing it. Started getting worse, but still not that serious until about 3 months ago, when we started seeking some help. Starting last week, we?ve run into a desperate situation where most of the time we are not able to get what done what we need to get done. This includes during the day, when most people are doing interactive processes and some are running ?batch? processes, and at night, when there will be 4-5 batch processes trying to run. But still, even since we?ve hit this bad situation, we have periodic times when the system runs perfectly, and last Thursday night and Friday was an example. For no apparent reason that we can find, the overnight processes flew through, and the system ran great on Friday until near the end of the day. There is definitely some correlation between the number of batch processes that are running at a given time and performance, but it?s been difficult to pin down a threshold or particular culprit. Again, it?s very inconsistent. Also, looking at the network and server stats, everything seems fine, though lately I?ve noticed that average disk queue length is spiking to 100 and staying there for long periods of time. But even that doesn?t seem to be a perfect barometer of performance. When it slows down, everything in Arev slows down, even bringing up a TCL or opening a program. However, the rest of the network is fine. Most people are also running Outlook, Office, and a few other applications. System: We have about 30-35 users on the system at peak time. We?ve definitely been adding a lot of data to our system over the past two years, and our entire database is about 6-7 gig. Network: 100M network, though some workstations still have 10M cards Server: NT4 server SP6a. We converted from a Novell server about 2.5 years ago. We still have Novell on our network for printing out of Arev. The server?s drives have about 11G on them, and about 25G free. I recently defragged the drive as part of trying to fix our problem. PIII 500MHZ CPU 512MB RAM System drive is on it?s own drive AREV is spanned on 3 drives in a software RAID 3 array 100Mbps full duplex NIC NT4 server SP6a No screen saver. TCP/IP is only protocol. Nothing else but AREV runs on this server. I don?t know of any unusual services, but if you?d like I can give you a complete list of all the services that are started on the server. Workstations: Mostly Win98, some Windows 2000 TCP/IP and IPX/SPX CPUs vary from 350MHz to 2GHz RAM varies from 64MB to 768MB Here?s a printout of what comes up when I do a ?who? command: Serial Number = A111561 Release Version = 3.1 User name = EAK Application name = DATA_1 Restriction level = Active Language Set = DEFAULT CPU Type = 80486 Math processor = 80487 Available memory = 322044 Descriptors used = 1490 Expanded Memory = Active NETWORK INFORMATION Vendor/Type = All Networks Driver 1.5.0.0 Maximum users = 42 Station ID = 0*ERICK OPTIONAL MODULE INFORMATION Name Release Version Date Installed ASCII 3.1 07:24:05 12 AUG 1993 dBASE 3.1 07:24:05 12 AUG 1993 IPX-NETWORK-DRI 1.12 09:52:38 03 MAR 1997 UPGRADE INFORMATION From Level To Level Date Installed 1. 2.0 15:23:00 14 APR 1990 2.0 2.03 18:33:24 11 DEC 1990 2.0 2.1 22:44:10 14 MAY 1991 2.1 2.12 09:38:38 21 OCT 1992 2.12 3.0 12:31:41 10 JUL 1993 EXPANDED MEMORY INFORMATION Expanded Memory Statistics EM Used = 512,000 EM Allocated = 4,194,304 Memory Region Segment Address 16K EM Window = DC00 (880K) 64K Overflow Buffer 1 = CC00 (816K) 64K Overflow Buffer 2 = 28F7 (164K) ---- === At 12 FEB 2003 04:32PM Richard Bright wrote: === Sig, Is he running Exchange Mail Server or other such stuff on NT Server - just wondering about RAM etc and interaction with NT Service. Richard ---- === At 12 FEB 2003 08:58PM dsig@sigafoos.org wrote: === Richard, Thanks for the questions .. According to his system guy nothing else is running on this server except the Arev service. Memory .. 512 .. 30-40 gig of free disk space dsig@sigafoos.org onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)" David Tod Sigafoos ~ SigSolutions Phone: 971-570-2005 OS: Win2k sp2 (5.00.2195) OI: 4.1.1 PII 300 laptop ---- === At 13 FEB 2003 04:16AM Richard Bright wrote: === The system as described should never bee under pressure. The only other off-beat things I can think of might be a) some MS housekeeping activity eg a quickfind index stuff or defrag/ virus scan cutting in on a shedule. b) a net card jabbering or similar flooding of the network. (Remember many PC's, though 'off' may be live to network.) Also if one runs Ghost to do image backups, this brings network to a grinding halt - even if particular server not involved - thus merit in Vlans and managed switches. Richard ---- === At 13 FEB 2003 08:30AM Don Miller - C3 Inc. wrote: === Just a thought .. I've seen a dedicated indexing workstation bog down a server when a lot of index transactions are being maintained from "batch" processes (particularly add/deletes). Also, have you checked the number of pending "lock" transactions? This sometimes manifests itself as a large amount of disk activity, too. Don M. ---- === At 13 FEB 2003 09:03AM dsig@sigafoos.org wrote: === Where do you check the locks? (i feel like the light is going out ) dsig@sigafoos.org onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)" David Tod Sigafoos ~ SigSolutions Phone: 971-570-2005 OS: Win2k sp2 (5.00.2195) OI: 4.1.2 ---- === At 13 FEB 2003 09:36AM Don Miller - C3 Inc. wrote: === Doesn't the server console maintain a table of locks? I may be off base here. I know I can do it in Novell and NT 4.x. Don M. ---- === At 13 FEB 2003 09:41AM John Bouley wrote: === You can do it with W2k also... Right Click on My Computer. Select Manage. Expand Shared Folders. Select Open files. HTH, John Bouley ---- === At 13 FEB 2003 10:51AM Dave Harmacek wrote: === To suggest another track: Do they have antivirus software running on the server? I just installed McAfee NetShield on a Windows 2000 server and found that ".OV?" are considered program files. I changed that entry to be just ".OVL" so that our Linear Hash overflo files wouldn't be examined. Dave ---- === At 13 FEB 2003 11:39AM Don Miller - C3 Inc. wrote: === Also true for Norton Antivirus products (server or not). Also InocuLan used to have the same problem, too. Don M. ---- === At 13 FEB 2003 01:03PM Pat McNerthney wrote: === Can't you just use the Task Manager to see what process is sucking on the cpu? Pat ---- === At 14 FEB 2003 09:03AM dsig@sigafoos.org wrote: === When the system is running fast the service is up in percentage. When the Avg Disk Que climbs and the system slows then the Fileservice drops. This tells me that it *shouldn't* be the service. It is not getting enought work to do so it drops. There is nothing else running on this machine except the Arev Fileservice and the 20-30 things that Microsoft, being the lean mean fighting machine it is, must have running. We do have, as I mentioned before, 5-6 files with are in overflow (more ov than LK) and of those there are 2 that are severe. As we know this will slow normal performance. But should not cause the Fileservice to go into a tizzy .. We are going to work on rebuilding these files but first are trying to find the main problem. At this point my vote is for Hardware. Either one of the drives are not working correctly, controller or ... Since they are stripped if one of the drives is not optimum then I do believe this can become a problem. I believe he is going to try running the system on another box. Not a major server but just a good box with memory and disk. This will tell us, after running a while, if it is the hardware or not. At least whether it is the server or not. dsig@sigafoos.org onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)" David Tod Sigafoos ~ SigSolutions Phone: 971-570-2005 OS: Win2k sp2 (5.00.2195) OI: 4.1.2 ---- === At 14 FEB 2003 02:38PM Gray Cunningham wrote: === There are a couple of very large files that are in desperate need of resizing, but they have been like that for years and the slowdown problem has only recently manifested itself. Their network guy seems to have improved the speed by adding a separate router for the printers...at least that's what I was told, but that info was given to me by a third party. They seem to be satisfied with the speed at the present time. ---- === At 14 FEB 2003 07:17PM Richard Bright wrote: === Based on what has gone before, seems most likely. People seem keen to go for the dual stripped (particularly IDE). At the server I would not be happy unless it is Raid 5 SCSI. Talking about slowdown on systems, recently I have had two clients in which PC performance dived. In the first I was able to demostrate HD going - but only by a failure to complete manufacturers standard tests in time allowed. Replaced drive and disk copied (we dont have a registry/software issue) --- performance zooomm! With other PC, IDE HD has passed all error test but I still think it has gone bad. Next step is to do disk copy, reformat and replace. Bet it is the disk! Richard Bright BrightIdeas new Zealand [[https://www.revelation.com/revweb/oecgi4p.php/O4W_HANDOFF?DESTN=O4W_RUN_FORM&INQID=WORKS_READ&SUMMARY=1&KEY=C209898741DB7B2185256CCB005BFB27|View this thread on the Works forum...]]