Re: HELP : S.E (7.x) nearly faster than OWS (7.x)

David Williams <djw@smooth1.demon.co.uk> wrote: >In article <6eiat1$qq1@camel15.mindspring.com>, Noble Systems Corp. ><noblesys@atl.mindspring.com> writes >>My experience with SE is that it is fast, *extremely* easy to >>administer, and totallly reliable. >> >>So, what's wrong with that? Nothing. SE RULES!!!! >> > Try bchecking your database. OK. We do that regularly from within cron just for a check (bcheck -n). Only *very* rarely do we encounter any problem, and that's easily fixed with a bcheck -y. > We have one customer who has had 4 crashes in the last 6 months. Sounds like they've got a problem. > We generally have to rebuild indexes on 4 tables >1.5Gb per crash. > > i.e. 1.5-2 days when users can't use the system How often? If it's the system mentioned above, I'm not surprised. > So much for a mission critical 24x7 system. Well, only a moron would run a mission-critical 24x7 system on SE! How would you back it up? But, despite what you may hear, not every system in the world is a "mission critical 24x7" system. Let's be real. > Of course Online would do fast recovery and stay consistent. Probably. > Or try a 4GL program > > DATABASE abc > > MAIN > UPDATE table1 set col1=1 > UPDATE table2 set col2=1 > WHILE 1=1 > BEIGN WORK > UPDATE table1 set col1=col1+1 > SLEEP 4 > UPDATE table2 set col2=col2+1 > COMMIT WORK > END MAIN > Whilst it is asleep kill -9 the sqlexec Excuse me, but I have seen *many* threads in this newsgroup regarding the fire-and-brimstone wrath-of-God problems that can occur from kill -9 ing (is that a word?) an sqlturbo (or equivalent) process. Why should SE be any different? > then > a) bcheck both tables > b) check the values in both tables, bet they are not the same!! If I ran a kill -9 against an sqlturbo doing the same thing, could you guarantee that they would be correct? For all versions of OnLine? > So much for transactional consistentcy. True. > PS We had another site where (as root they did) > X=`ps -ef | grep ?` > kill -9 $X > Of course someone create a file in / called q > Hence this became > X=`ps -ef | grep q` > kill -9 $X > Hence ALL sqlexec's recieved a kill -9 and we had 5 database > corruptions in a row...2 in the same day.. Yep. That sucks all right. Just goes to show you that you can make it foolproof, but not idiot-proof. > SE f**ks up on a regular basis for us and of our 15 sites we usually > bchekc -y one every 2-3 weeks at least. With our 3 online sites we > have done no fixes in the last 18 months despite You've got some serious problems. We're supporting 250 sites with well over half of them 2.10 and we don't have that many problems. > a) Site A having complete power failures (without UPS's) 3 times > in 2 weeks. They logged no support calls and have no application > errors on online.log errors. Just the expected fast recovery > and this in Online 5!! > b) Site b crashin Online twice in a year (OWS 7.20). > None of our SE sites have lasted a year without at least one > bcheck -y or drop/rebuild indexes. That's odd. > I HATE AND LOATHE SE, I'VE LOST AR LEAST 5 NIGHT UP TO 1AM OF MY OWN > ^^^ > TIME, UNPAID JUST BECAUSE SE CAN't ROLLBACK UNCOMMITED TRANSACTIONS > ^^^^ ^^^^^^ > WHEN IT STARTS.. HEAP OF SH*T!!! > Sorry but that is just how I feel. The sooner we get our sites onto > Online 7.23 on Sun boxes running Solaris 2.5.1/2.6 the happier I will > be. Well, if that's been your experience, I can see why you'd be upset. My experience is that SE is a good, viable solution for many sites. For us, it's been fast and solid. > >-- >David Williams >Maintainer of the Informix FAQ > Primary site (Beta Version) http://www.smooth1.demon.co.uk > Official site http://www.iiug.org/techinfo/faq/faq_top.html >I see you standin', Standin' on your own, It's such a lonely place for you, For >you to be If you need a shoulder, Or if you need a friend, I'll be here >standing, Until the bitter end... >So don't chastise me Or think I, I mean you harm... >All I ever wanted Was for you To know that I care