Re: Online Archiving - Not ?

> In article <1qmunqINNcko@emory.mathcs.emory.edu> I wrote: > : > : Here's a kettle of fish; > : > : Online 5.00 UC2 > : 4gl 4.10 UD2 > : HPUX 8.06 > : > : We have a home rolled transaction processor which is continually reading the > : contents of a table, deleting the rows and processing the data found therein > : and writing it to the appropriate places. It is processing at a rate of > : 3-4 transactions/sec normally and when catching up to backloads - at about > : 100/sec. > : > : During an online archive, which lasts several hours, everything proceeds > : smoothly until suddenly all of the users are locked up for 5-10 min. After > : that things catch up and everybody is happy. > : > : What is happening to cause the lock-out and how do we work around it? Thank you all for the responses. The logical logs filling is not a problem since we don't have logging turned on. We go through about 1 logical log per day. My GUESS is that the Online allows you to continue writing transactions during an archive until a checkpoint is required at which point it needs to synch the archive with the checkpoint. The lock out would then occur while every- body who's working goes into a checkpoint_wait status until a convenient moment can be found by tbtape to stop for a second. OR - and it's still the same thing - the checkpoint fires and cannot complete while a couple of pages to some table or other are being archived. I don't see an easy solution - maybe going to incremental archives, which would make the whole archive shorter, might lower the contention. j. _____________________________________________________________________________ Jack Parker - Contractor | Hewlett Packard, BSMC Boise, Idaho, USA| If you keep staring at it like that, jparker@hpbs2561.boi.hp.com | your nose is going to grow (208) 323-5388 (W) (208) 384-1623 (H) | into the bark. ^^^ 396 as of 6:30am April 29th _____________________________________________________________________________