Reducing number of updates being done to Oracle data
- Article Type: General
- Product: Aleph
- Product Version: 20
Description:
Do you know of anything I could disable that would help keep our /oradata/aleph20/ from filling up so fast?
It seems to be filling up rather quick today for some reason so I wasn't sure if there was a journaling feature or something that we don't really need to be running that might help.
Resolution:
One of the main sources of large numbers of updates are the ue_nn daemons. (See KB's
8192-7543, 8192-10426, 16384-28562, and 6658 in this regard.)
I see that the daemons (such as, sys01 ue_01) have not been restarted for 6 weeks.
I suggest that you stop and restart the sys01 ue_01 right now.
As described in KB 16384-21105, we suggest that aleph_shutdown/aleph_startup be run weekly.
Also:
I find that you have 179 million z0102 records. This makes the ue_01_z0102 process do a lot of work when you update records.
Since you have 5.9 million bibs, I suggest (as discussed in KB 8192-6888) that you set col. 8 in tab_base.eng to "N" for any bases which have more than 600,000 records. That would be all of these (as seen in util h/1):
USD | 644943| Yes | Yes |
USDIS | 645312| Yes | Yes |
USDVE | 677235| Yes | Yes |
USDVI | 677604| Yes | Yes |
USDALL | 696998| Yes | Yes |
USDWE | 664706| Yes | Yes |
USDWI | 665075| Yes | Yes |
BRO | 783190| Yes | Yes |
SFL | 600903| Yes | Yes
VERMILLION | 766535| Yes | Yes
Eliminating the z0102 for these large bases should reduce the ue_01_z0102 load considerably.
- Article last edited: 10/8/2013