I understand that this subject has been discussed several times over quite a long time period. After viewing all of the archives, I'm hoping to have a few questions answered on this topic I did not find in the archives.
Our company is a wholesale distributor running FoxPro 2.6 DOS on Novell 4.10. For quite some time we have been running Win95/98 workstations on the network with the Novell32 client w/out any database/index corruption. In December of last year we upgraded our Pentium 266MMx PCI/EISA bus server to a Dell PowerEdge 2300 server. The previous server was built in-house, used a PCI RAID system by AIWA, and EISA dual channel 10mbit network cards from SMC. The Dell is also using PCI disk subsystem (not RAID), however the NIC cards are SMC 10/100 PCI dual channel. Why am I going into all this detail? Well the corruption problem developed 1 week after installing the new Dell server - nothing else changed.
As a result, we researched and discovered the Client32 issues and reloaded every Win95/98 workstation (from scratch) with the Microsoft Netware Client to insure that was not the problem. After exploring other internal possibilities we have opened an incident with Novell and have not discovered the source of the problem yet. We are sure of this: it can't be the Novell 32bit client - we aren't using it! On the other hand, after reading other posts, comparing to our situation, I have a few questions for those of you who have experienced this.
1. What is/was the exact nature of your "corruption"? Ours is that the data is being shifted to the right an arbitrary number of bytes at an arbitrary point in the record causing the remaining field to contain invalid data. On occasion this "wraps" to the next record.
2. What hardware are you running? Did your problems start occuring after a hardware change? I find it coincidental that ours developed as soon as we did added 3 components to the network: 1)PCI NIC on the server, 2) a mixed 10/100 environment vs. strictly 10mbit, and 3) Pentium II servers and workstations. Is it possible that any such changes were made within a few days of when your problems developed?
As mentioned, we have been working with Novell for some time now and have checked/changed MANY things that could be the cause - from clients to specific versions of NLM's, HAM's and all kinds of stuff. At this point I'm ready to blame hardware and I'm hoping we might find a common thread here with hardware - PCI/Fast Ethernet/PII etc. Thank you for reading this long post and for any specifics you might be able to contribute that will help us troubleshoot the "real" cause of this "corruption".
Joe Miller
jmill@ezo.net
