Dear Olaf, Many thanks for your excellent analysis.
> >> My primary concern now is to prevent a dead-lock. > That seems to make sense now (I assume you're working > "near a deadlock" with your multipe-client-requests, not > going to sleep properly before the next retry). Still makes no sense to me, how the absence of re-try code, while very silly - yes;), can explain the vast difference in performance I see between the "local-disk" scenario and the over-NFS scenario? In other words can you explain the drop in performance by the absence of re-try code? Even prior to eventually hitting a dead lock the performance over NFS is lousy? Yet you say, it shouldn't be that bad due to the (relatively) small size of the database and updates? The performance in the local-disk/concurrent scenario is excellent... ? Befuddled, -rosemary. > Sorry, but cannot help with Code-Snippets, I'm not a > C-Coder - but we start with "Begin Immediate" in our > Write-attempts - in case of busy then a few rounds with > shorter sleep-intervals - then looping with a longer one, > but then after a while with again more and more shortened > sleep-intervals (to prevent starvation) - more or less the > same wait-strategy then in the readers. > > Maybe just post your code - how you initiate your writes > and how you've implemented your current wait-loops > (regarding the timeout-handling) - I'm sure the cracks here > will find what's wrong. > > Olaf > > > > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users