This content has been marked as final. Show 21 replies
If itis just a message it should be rmoved by Oracle as this is in the alert logs and is supposed to worn you about errors not to waste your time.
this message can be ignored , if you are not experiencing a redo contention problem
You need to understand that having so many redo log members across just two of your mount points (which is also used by the DB Writer for data, temp, etc., ) can slow down your log. I've faced similar situation, where if the data /index/temp mount points are busy and using up most of it's bandwidth, the log writer will suffer if they're also using the same mount points. Plus I don't see any reason why you should have that many redo log groups. Even if it's required, and claim that your database large and extremely busy, the number of mountpoints (just 2) is not sufficient to distribute your I/O load. You need to further separate your OS, /tmp locations and move them to alternate mount points.
My suggestion is to first see how busy your I/O is during peak times and what % of the I/O bandwidth is being used. Are you getting I/O cache saturation at this point? If so during these times, your redo logs will become much slower. Not to mention you've so many of them which only could make things worser.
Add one or two separate (physically separate) mountpoints, and direct your redo logs there, if you see your I/O is being pounded by dbwriter.
Easy way is to use dbconsole/grid to get to the performance tab and check your I/O usage (MB/sec) and compare that with the number of usable spindles in your RAID or sustained max I/O performance and make sure your i/o is the bottleneck (I very surely think so).
Everson Piza started this thread six years ago & has not posted here in over a year.
I doubt Everson Piza will benefit from your words of wisdom.
In the future consider responding to more current postings.
I wonder why no one ever asked how often the logs are switching?
And what sb said.
iam also getting the same problem ie,(Alert.log: Private strand flush not complete ) and i have made many changes by increasing redo size,dbw,log buffer but it didn't work
here for my all datafile auto extend is ON,
and all datafile are nearly 85% filled,
in my case do i really need to increase the size of datafile(even auto extend is ON)
See MOS Alert Log Messages: Private Strand Flush Not Complete [ID 372557.1] and don't worry your pretty little head about it.
In the future if you want to reference an old thread, please start a new one and put a link to the old one in it. And search for the answer before you post, you might be able to figure it out yourself. You are expected to do that, you know.