Prev: IBM DB2 Database Specialist - Saudi Arabia
Next: 100% without investment online part time jobs..(adsense,datawork,neobux..more jobs)
From: shorti on 4 Aug 2010 19:20 I am using DB2V9.5 on AIX6.1. I have a machine that the database backups are now over 3 times as large as it should be. Normally, a REORG on our systems cures the problem but this time it did not. I looked around on the systems and do not see anything that sticks out as being a problem (ie history files small, small number of active and archive logs, no pndfreepgs, etc). We have hundreds of systems and this is the only one Ive seen like this. We do a full online database backup once per hour. The backups for this size database (only 330k records on the largest table) should be less than 200MB in size but are now almost 700MB. The machine was brought down (instance stopped and started) and an offline reorg of the tables and the indexes (using indexall) was performed but it only reduced the database size by about 40MB. I then ran reorgchk (a week after the reorg was performed) and it is showing a few indexes needing reorg (not uncommon after a week). I then tried an INPLACE ALLOW WRITE reorg on the first index...after two hours I saw that I was barely 5% through the count for that one index. Since the customer was running at the time I stopped the reorg and waited for a more idle time (even though it didnt look like the machine was running very heavily). 7 hours later I found an idle window and retried the REORG again..this time I did the table first (still using INPLACE because the machine could not be taken offline at this time). The table reorg finished within a few minutes. I then tried that same index again. It ran for about 3 hours and was only about 1/3 way through before my idle window expired...so I had to stop it again. With those calculations, it looked like it would have taken 9 hours to do that one index!! Again, this machine had hardly anything going on and typically on our busy machines with much larger databases it would only have taken 3-4 hours max to reorg this particular index. Now I am thinking something is very wrong with some of the indexes. Has anyone experienced something like this before? Is there something else I can look at that might indicate what is being backed up that is making it so large? Also, is the slow index reorg an indication of anything? We have planned an outage with this customer to again try an offline reorg but I am afraid it wont cure the problem since it didnt do anything the last time. Please let me know if I am missing some key item. Your help is much appreciated! |