What causes GFE code 8 row terminator missing? (AREV Specific)
At 15 JUL 2005 03:59:28AM Janet S Scott wrote:
Hi All
One of our customer sites has AREV 3.12, running on an NT/windows 2000 network and XP workstation clients. They have all networks driver 2.1 I think.
They are getting GFE Code 8 row terminator missing at the rate of about 2 a week. Doing the dump thing seems to fix them but I think they're finding it annoying. The problem seems to be occuring over several different tables. I'm not sure but I think they are the larger more frequently updated tables.
Until recently it seemed their all networks driver was not installed correctly. When you did a "who" it showed "Unknown driver". There were also some problems where the network administrators had "fixed" a few things like turning the virus checking back onto the arev files without checking with people who know what they're doing.
Also most of the PCs do not have "write behind caching" turned off. I have asked for this to be done but on a "as needs" basis ie if someone gets the GFE at their workstation - fix it.
I suspect that as they get new staff, the newbies have a habit of crashing out of the arev system either by turning the PC off or forcing the dos window to close without exiting the application gracefully - I think this doesn't help.
If anyone has any other ideas about what causes the row terminator missing thing (or how to make it stop), can they please post an answer in this thread. Thanks.
I am wondering, too, if rebuilding the problem tables (could take the system down for hours) would help?
Janet
At 15 JUL 2005 07:46AM support@sprezzatura.com wrote:
Janet
Until every workstation is configured correctly (write behind caching off etc) then there is little point in speculating on cause. Suffice to say that in a correctly configured network using a network product you shouldn't be seeing GFEs on a regular basis.
PS Hiya!
support@sprezzatura.com
The Sprezzatura Group Web Site
World Leaders in all things RevSoft
At 15 JUL 2005 03:18PM Victor Engel wrote:
The only thing I would add to what Sprez. posted is that just because a particular workstation is not exhibiting a problem doesn't mean it's not part of the problem. It could very well be that the workstation not showing symptoms is fouling things up for everyone else.
If I had any sort of authority at this location, "as needed" would mean, "Is Arev being used on the workstation? Then set things up to spec." Period. No exceptions. The weakest link in the chain will foul things up for everyone else.
Suppose, for example, you are using the NLM. The NLM performs its own caching on the server. Now if a workstation does some table updates without going through the NLM, then the file is updated, and everything looks OK from the perspective of that workstation. However, since the transaction did not go through the NLM, the cache on the NLM did not get updated. Now, when another user, configured properly, performs an operation on the same table, the transaction will go through the NLM, and, possibly, data in the NLM's cache will be used to satisfy the request. This will be in contradiction to what is on file, possibly revealing itself as a GFE, for example. In this scenario, there isn't even an actual GFE present on disk, but users correctly configured exhibit the GFE, whereas the user incorrectly configured does not.
Victor