>>>>In a Web environment, I have 4 VFP servers running to process the incoming requests. I have found that occasionnaly, one save by a VFP server is not reflected immediately on another VFP server. What I mean by that is that it may takes up to two seconds before the other servers can read the updated data from the tables.
>>>>
>>>>It seems that I have to force a flush here.
>>>>
>>>>What would be the best approach to make sure all data is saved on disk immediately after the VFP server that did the save transaction?
>>>
>>>There was discussion about a similar problem but programs modifying table were located on separate PC's. If I remember correctly the solution was, in addition to what VLAD suggested, to RLOCK(myalias,0) to lock table header before read to make sure that updates from another program are visible.
>>
>>This is a strange situation. As I mentioned to Vlad, I am wondering if we force to use SYS(1104) in such an environment.
>
>
>All Vlad sugestions work on VFP level but if delay is reletaded to OS than they wouldn't help.
>Small correction, after RLOCK(myalias,0) you should immediatelly do UNLOCK.

We can't afford to apply a development rule to do that on every save we do. We are looking for an approach at the top level just before returning the HTML code to the server. I will verify for the SET REFRESH. Maybe, as oppose to what I think, it is not set to 0.