>>I have sometimes seen network switchgear go wonky if a site has power problems. You could try power-cycling the switch(es), especially if they haven't been power cycled for a long time.
>>
>>I've seen network delays/disconnects caused by bad patch cables but that only affects one workstation at a time. Server and closet patch cables are generally OK as long as they're good quality cables that meet spec e.g. Cat5e or better for gigabit networking. If they haven't been disconnected for years it might be worth disconnecting, then reconnecting them in case electrical contacts have become oxidized.
>
>Even gold-plated can lose contact by having dust settling by vibration from the fans. Same cure applies.

After blowing out the dust :)

>>Exactly how are you rebuilding your indexes? Just running REINDEX is not good enough. One way to do it properly is to DELETE TAG ALL then re-add the tags. However as I understand it this can break referential integrity if you have RI defined in a VFP database. My preferred method is to overwrite the .CDX with a known good one (even if it's out of date, could be from your dev computer etc), then REINDEX (which will bring it back up-to-date).
>
>My preferred method is to generate the code to save all the RI stuff etc, drop it, then drop each index by stuffing a zero in 29th (or was it 28th... depends how you count) byte, physically erasing cdx files, then building indexes in (generated) code, then reapplying the RI stuff. The code is a bit convoluted and generally not used in the last 15 years, but if anyone wants the generator, it still works.

ISTR some people use improved RI code (Steve Sawyer?). Not sure how the RI in an existing DBC would be generated in that case.

I see no reason to mess with any RI code at all, if the task is simply to reindex tables. Do you know of any reason? There are no backlinks in index files, and a simple CDX overwrite with a known good one, plus REINDEX, has always worked well for me.