RE: Stupid DBA tricks

Set LBUPRESERVE to 1 in the onconfig flie. This should allow you to run your log backup to /dev/null. -----Original Message----- From: Michael Hoffman [mailto:mrh@panix.com] Sent: Monday, January 29, 2001 1:20 PM To: informix-list@iiug.org Subject: Stupid DBA tricks Hi All, So, I have a brand new system from my own "learning experience": a single CPU box with 2 4GB Hard drives and SUN OS 5.7 I'm running IDS 2000 9.21.UC1 in this test environment while preparing to migrate the "real" servers to IDS 2000 within the next few months. As I was copying the current database from the live servers to this test environment, I ran into a Long Transaction problem ("Load from ...." on a large table). Instead of using 'hpl' or even breaking this file into parts, I upped the LTXHWM and LTXEHWM to 70 and 80, respectively. Of course, I also forgot to set LBU-Preserve to 1. Needless to say, this time when the Long Transaction hit, it didn't bomb the load, but instead hung the DB. The online.log message read "Waiting for Next logical Log". Trouble with a Capital T! I couldn't force a checkpoint because the logs were totally full. I tried 'ontape -a' but got an error "LogBackup to /dev/null not allowed". I'm assuming that this is also because there were no free logical logs to write out the fact that an archive was being performed. Since this is "my" play server and the DB was just being built, I took the easy way out and ran an 'oninit -i' and started over from scratch. For my future knowledge and further education, should I ever decide to listen to the little voices in my head saying "why not try *this*?" and end up in this situation again, is there ANY way to bounce the database and *** BYPASS *** the Fast Recovery step? Or was my "easy way out" really my only choice anyway? Thanks, Michael Hoffman

Hi John, but the logical log backup will be the last action on that server :-)) John Carlson wrote: > > Set LBUPRESERVE to 1 in the onconfig flie. This should allow you to run > your log backup to /dev/null. > > -----Original Message----- > From: Michael Hoffman [mailto:mrh@panix.com] > Sent: Monday, January 29, 2001 1:20 PM > To: informix-list@iiug.org > Subject: Stupid DBA tricks > > Hi All, > So, I have a brand new system from my own "learning experience": a > single CPU box with 2 4GB Hard drives and SUN OS 5.7 I'm running IDS 2000 > 9.21.UC1 in this test environment while preparing to migrate the "real" > servers to IDS 2000 within the next few months. > > As I was copying the current database from the live servers to this > test environment, I ran into a Long Transaction problem ("Load from ...." on > a large table). Instead of using 'hpl' or even breaking this file into > parts, > I upped the LTXHWM and LTXEHWM to 70 and 80, respectively. Of course, I > also > forgot to set LBU-Preserve to 1. > > Needless to say, this time when the Long Transaction hit, it didn't > bomb the load, but instead hung the DB. The online.log message read > "Waiting > for Next logical Log". Trouble with a Capital T! I couldn't force a > checkpoint because the logs were totally full. I tried 'ontape -a' but got > an > error "LogBackup to /dev/null not allowed". I'm assuming that this is also > because there were no free logical logs to write out the fact that an > archive > was being performed. > > Since this is "my" play server and the DB was just being built, I > took > the easy way out and ran an 'oninit -i' and started over from scratch. > > For my future knowledge and further education, should I ever decide > to > listen to the little voices in my head saying "why not try *this*?" and end > up in this situation again, is there ANY way to bounce the database and > *** BYPASS *** the Fast Recovery step? Or was my "easy way out" really my > only choice anyway? > > Thanks, > Michael Hoffman -- Stefan Weideneder Phone: +49 89/3565478-2 --------------- --- Fax: +49 89/3565478-3 ------------- ------ mailto:/stefan@weideneder.de --- -------- http://www.weideneder.de -----