Re: Why a long transaction cannot rollback?? (2)

In article <Ak6nSiAfxqTzEwvk@smooth1.demon.co.uk>, David Williams <djw@smooth1.demon.co.uk> wrote: >In article <5ii2dl$4ud$1@gulfa.kuwait.net>, Rudy Fernandes ><rferdy@kuwait.net> writes >> >>This is correct,too. On testing, I've found that OL uses Log Ids to >>detect HWM and EHWM. With 6 logs in my test instance, OL invariably >>aborted when the transaction moved into the x+4 log. It did not >>matter whether the Tx started at 5% of log x or 95% of log x. >> > No way - I always assumed that it used sum (pages used)/ > sum(pages) for all logical logs. Does this mean that with 3 logical > logs that it only checks at 33%,66%and 100%? SO that with 4 logical > logs changing LTXHM from 40%to50% has no effect??? > I'm afraid that LTXHWM checking does work the way I mentioned (v7.2!). In fact, I decided to test the 3 log case (Madeline, I hope you are enjoying all this perspiration :)), and I found the following: Testing with LTXHWM at 34 first, OL behaved 'predictably' - when it switched to the x+1 log, it aborted, irrespective of where in x it had started. (In some cases the tx inserted 25 records; in others 200!) With LTXHWM at 30, the 'long' transaction again aborted at the point where OL switched to x+1, indicating the redundancy of a value of LTXHWM below 33 in a 3 log situation (as OL checks LTXHWM only at log switch). Finally, with LTXHWM at 70, OL went smoothly into log x+1 (I took it to close to 60% before aborting to avoid the 'no free log lock'. This is a purely theoretical exercise for my site where our live instance has 100 logs. But, for sites which maintain fewer than 10 logs, be aware! ---------------------- Rudy Fernandes GIC, Kuwait OL 7.20, 4Gl 6.04 ----------------------