Re: HELP!!!!

Can this problem can be avoided? Would more memory and drive space prevent Informix from "crashing"? This seems like a very bad thing for a DB to do... I would go to upper management and explain that more $$$ needs to be invested in hardware!!! Steve -- Ignorance can be cured! scooper@thecompounders.com Nigel Gall <nigelg@ppsl.com> wrote in article <5ua396$11q@cssun.mathcs.emory.edu>... > > Hi! > > Sorry to be the one to tell you this, but there are only two things you can > do: > > 1: If you have the 24x7 class of support from Informix, you may contact > them and they will use some special tool to manually force some flag (I > think it has to do with the checkpoint) to true, so that your system won't > spend so much time in fast recovery; the downside of this is that your > database will be left in an inconsistent state since your transaction which > caused the lock-up (your long transaction) will be incomplete and not > rolled back. > > 2: Restore from backups. > > I had to do this a few weeks ago, and I chose to restore from backup > because I don't have the 24x7 support option as well as I cannot accept > having my database in an inconsistent state. > > When you finish restoring from backup, there are a few things you can so to > prevent this from occuring again: > > 1: Lower your LTXHWM and LTXEHWM to low values in the range 20 to 35. > > 2: Increase the number of logical logs you have, if possible. > > 3: Enforce a rule that your users *MUST* inform you when they are going to > run any program which may cause a lot of updates, such as a posting program > that covers a number of months of data, or an end of year posting. That > way you may be able to run system checks while their program is in > execution and you'll be able to take evasive action before OnLine crashes. > > 4: If the program is an inhouse-built program, ensure that transactions are > committed or rolled back with a smaller amount of changes; in other words, > don't let your programs/routines make modifications to a large number of > rows in a single transaction (this fills the logs quickly). > > 5: If a long transaction is in progress, and your logs are approaching 50% > full (with your LTXHWM and LTXEHWM values being less than 30), shutdown > OnLine using the Immediate option; this will kill all server processes and > bring OnLine to quiescent mode. OnLine will then rollback your long > transaction (as well as any other open transactions that were in execution > at the time; so be prepared to explain your actions to some angry users; if > possible, let the extension number of the user responsible for the long > transaction be accidentally leaked to all angry users :). This is a > brute-force method of dealing with long transactions. If you have to do > this and your users are angry, tell them that the only other alternative > was to let the system crash and restore from last night's backup, resulting > in them losing a wole day's worth of work rather than their current open > transactions (which may be small and easily re-entered in the system). > > Sorry I couldn't be of more help, but this is a hard lesson we all go > through at some point. Your main responsibility now is to recover and > learn from it. :) > > At 08:15 PM 8/29/97 EDT, you wrote: > > > >I am having problems with Informix OnLine.. Please if someone knows > >All my logical logs are full because a long transaction... > > (the device for backup is in /dev/null) > > > >I have many databases and I don=B4t know how to put the online in Online= > > Mode.. > > > > Maybe someone knows hoy can I ingore the fast recovery process or wipe the > >logical logs... I will apreciate > > > >} Goldy Pastor > >} La Paz - Bolivia > > > > > > > > > >} > >} HELP!!!!!!!! > >} > >} > >} I need help.. I have the following message.. and the OnLine.. ABORT. > >} How can I put in OnLine... > >} > >} > >} > >} Fri Aug 29 12:20:39 1997 > >} > >} 12:20:39 INFORMIX-OnLine Initialized -- Shared Memory Initialized > >} 12:20:39 Physical Recovery Started > >} 12:20:39 Checkpoint Completed > >} 12:20:39 Physical Recovery Complete: 0 Pages Restored > >} 12:20:39 Checkpoint Completed > >} 12:20:39 ERROR: logundo(0) iserrno 0 us 0x803678 pid 0 > >} tx 0x809690 loguniq 1976 logpos 0x690794 > >} > >} 12:20:39 INFORMIX-OnLine Must ABORT > >} Log Error 'rollback() - logundo() FAILED' > >} us 0x803678 pid 0 us_flags 0x101 > >} tx 0x809690 tx_flags 0x82403 tx_loguniq 1976 tx_logpos 0x690794 > >} 12:20:39 -- Online Aborting -- us=803678, pid=678, uid=12 > >} > >} > >} > >} Thanks... > >} > >} Goldy Pastor... > >} > > Best regards, > Nigel > +-------------------------------------------------------------+ > |Name : Edmund Nigel Gall Tel: (868) 636 3153 | > |Title : Information Systems Specialist Fax: (868) 679 3770 | > |Company: Process Plant Services Limited | > |Address: Atlantic Avenue, Point Lisas Industrial Estate | > | Point Lisas, Couva, Trinidad & Tobago, W.I. | > +----- mailto:nigelg@ppsl.com ------ http://www.ppsl.com -----+ > >