Problem with database back up

aleks

New Member
Hello!

I have a problem withh progress database backup.
Usually back up log write a state every 10 seconds. But today it writes 10 strings, then waiting something for around 15 minutes, and then continue writing another 10 strings, and then another pause.

Usually database backups about 1h, but today it's 9h.

What is it? What to do?

Example of back up log file:
Backed up 6924779 db blocks in 02:28:59
Backed up 7004903 db blocks in 02:29:09
Backed up 7084824 db blocks in 02:29:19
Backed up 7164812 db blocks in 02:29:29
Backed up 7243782 db blocks in 02:29:39
Backed up 7321393 db blocks in 02:29:49
Backed up 7388338 db blocks in 02:29:59
Backed up 7435448 db blocks in 02:34:13
Backed up 7499499 db blocks in 02:34:23
Backed up 7578706 db blocks in 02:34:33
Backed up 7658933 db blocks in 02:34:43
Backed up 7738887 db blocks in 02:34:53
Backed up 7814935 db blocks in 02:35:03
Backed up 7886942 db blocks in 02:35:13
Backed up 7952767 db blocks in 02:35:23
Backed up 7969546 db blocks in 02:40:28
Backed up 8035031 db blocks in 02:40:38
 
  • What version of Progress / OpenEdge is this?
  • On what OS does it run?
  • Is this an on or offline backup - to you have a transaction load on the database while the backup is running?

Heavy Regards, RealHeavyDude.
 
1) OpenEdge Release 10.1C04 as of Fri May 29 21:42:49 EDT 2009
2) Sun Microsystems Inc. SunOS 5.10 Generic January 2005
3) Online backup, yes, we have transaction load during online backup
 
Congratulations - that is the same configuration we have in our production system. Just kidding ...

The transaction load during the online backup could be a perfect explanation for make the online backup slow. Usually the online backup goes through the database blocks sequentially and backs them up one by one. As soon as a update to a database block further down the road is pending, the online backup suspends the transaction and jumps to that black, backs it up. As soon as that back is backed up the transaction goes through and the online backup jumps back to where it was before it had to jump. If there is a heavy transaction load during the online backup it needs to jump back and forth all over the place, slowing the transaction and itself down considerably.

Are you aware of an unusual heavy transaction load while the online backup is running? Do you have any statistics on what usually is happening transaction-wise on the system at that time?
Just mentioning because it hurt us in the past: There is an option "forced direct io" on the mount for the file system which is not supported by Progress and which in case slowed down the database activity and the online back so that everything almost came to a halt. Switching it off brought back normal performance. It seems to be that there is a recommendation from Oracle to use it although it is discussed controversial and most Oracle DBAs are not happy with it. But that's how it sneaked into our system as Progress is a niche product in our company.

Heavy Regards, RealHeavyDude.
 
We are working with this configuration in same regime for around 3 months. Before we had no problems. Problems begins for a week ago. And today it became crytical.
 
Are you aware of an unusual heavy transaction load while the online backup is running? Do you have any statistics on what usually is happening transaction-wise on the system at that time?

These questions are the ones that you need to focus on answering IMO :)
 
Back
Top