I've been working with this app for several years and, in general there have been relatively few problems. We have an overnight maintenance routine that does a comprehensive cleanup on the files. With the odd exception, it runs pretty smoothly and if it doesn't there's usually an obvious reason why not. Last night the overnight maintenance failed to complete and this morning, I couldn't get a compact to complete normally. Eventually I traced that problem to a corrupt table, removed/replaced the table and the compact seemed to run.
I was looking at it again tonight and found that, although a compact did complete without errors (or so it seemed) when I looked in the Trace Error window, a huge number of the DDD files are listed in the in the error page - just the filename.ddd with no other information. I tested a couple of the files/tables listed with a dictionary.update() and that generated a trace window error as well. This is happening on both my development files at home as well as the production files. I have no idea how long this has been going on since I haven't been getting visible Trace errors.
Any of the tables that I go to can be reindexed, packed etc without incident and today's production performance seems to have been unaffected. All the tables and sets look normal, no outsized files and all the fpt files seem OK.
We are running an odd mix where physical desktop machines use remote desktop to log into virtual workstations which open a shadow app on that workstation that is networked to a virtual server. Seems complicated but has been running pretty well for a couple of years. We recently added some "regular" desktops which are shadowing the app directly from the virtual server.
Thinking aloud: it's possible that the client may have pushed a machine into use which is running a different build than I am here and that is running on the server. I'll check that tomorrow. Otherwise I'm at a loss. Has anyone else seen data dictionaries listed like that in the Trace Window? Any ideas?
I was looking at it again tonight and found that, although a compact did complete without errors (or so it seemed) when I looked in the Trace Error window, a huge number of the DDD files are listed in the in the error page - just the filename.ddd with no other information. I tested a couple of the files/tables listed with a dictionary.update() and that generated a trace window error as well. This is happening on both my development files at home as well as the production files. I have no idea how long this has been going on since I haven't been getting visible Trace errors.
Any of the tables that I go to can be reindexed, packed etc without incident and today's production performance seems to have been unaffected. All the tables and sets look normal, no outsized files and all the fpt files seem OK.
We are running an odd mix where physical desktop machines use remote desktop to log into virtual workstations which open a shadow app on that workstation that is networked to a virtual server. Seems complicated but has been running pretty well for a couple of years. We recently added some "regular" desktops which are shadowing the app directly from the virtual server.
Thinking aloud: it's possible that the client may have pushed a machine into use which is running a different build than I am here and that is running on the server. I'll check that tomorrow. Otherwise I'm at a loss. Has anyone else seen data dictionaries listed like that in the Trace Window? Any ideas?
Comment