Re: onmode -B

tomL wrote: > apparently, onmode -B is an undocumented feature that many people use. > > We are running informix 9.21 HC5-1 on UNIX-HP ver 11.00 and have > employed the contorted method of running a cron job that fires onmode > -B, sleeps for 2, then runs onmode -c. It has taken our long & painful > checkpoints that use to last anywhere from 20-132 seconds (!) and > dropped checkpoints to 1 second - a remarkable improvement. What I > would like to know is if anyone know if there is a price to pay for > doing this? The complexity of running the cron job - mainly. This technique does not clear the buffer pool completely (as Ravi - rkusenet - implied); it simply writes out all the dirty pages (marking them as clean). The pages in the buffer pool do not have to be re-read to be re-used. There is a small potential penalty; your users might modify a page between the time when the onmode -B flushes it and the time of the checkpoint proper; that page will be written twice in short succession, which would not have happened without the onmode -B. On the other hand, this minor penalty is usually more than offset by the reduced duration of the checkpoint proper. As Ravi also pointed out, there is a slight chance that something weird could happen and overfill the physical log (75%) or otherwise trigger a checkpoint. However, by setting the checkpoint interval to, say, 10 minutes and running the cron job every 5 minutes, and by sizing your system appropriately, you can pretty much avoid those problems. You might have problesm if the cron job stops working - but then the automatic systems would kick in and do checkpoints anyway, just more slowly. > We currently have a case in Informix support and have not heard > anything back yet so we are going with this "solution" for now because > it is really needed. Go with what works for you. -- Jonathan Leffler #include <disclaimer.h> Email: jleffler@earthlink.net, jleffler@us.ibm.com Guardian of DBD::Informix v2003.04 -- http://dbi.perl.org/