1

Topic: High waitings log file sync

Kind time of days!
DBMS: Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production
I see high waitings log file sync (the screen see)
But thus disks on which lie redolog on record and on reading have the response to 3 msec (reading 3 msec, record 1,5 msec). Like at such responses it is not necessary to reflect on transfer Redo log on other disks?
The second direction I see on the hottest tables on record removal few used indexes.
Generally such values log file sync it is a problem?

2

Re: High waitings log file sync

And what switching of logs at an o'clock? Probably small size REDO

3

Re: High waitings log file sync

app.and;
Can frequent commit at you.

4

Re: High waitings log file sync

app.and;
What processor?

5

Re: High waitings log file sync

Andrey Panfilov;
And here the processor?

6

Re: High waitings log file sync

BigBudda wrote:

And what switching of logs at an o'clock? Probably small size REDO

5 files on 8 Gb everyone

7

Re: High waitings log file sync

Time of switching RedoLog of 7 minutes

8

Re: High waitings log file sync

app.and wrote:

Andrey Panfilov;
And here the processor?

Because on any shit of type Sun Niagara with frequency of a kernel already 300Mhz high waitings log file sync quite to itself the "normal" phenomenon

9

Re: High waitings log file sync

app.and wrote:

Time of switching RedoLog of 7 minutes

And how many dens in state ACTIVE?

10

Re: High waitings log file sync

Vyacheslav Ljubomudrov,

11

Re: High waitings log file sync

Well also what under the same report with ?
Values of parameters COMMIT_LOGGING, COMMIT_WAIT?
Yes the report here is the most easier to lay out, and that is not present any information

12

Re: High waitings log file sync

Yes is not present here any  log file sync. 3ms this quite normal time.  on a disk it log file parallel write, instead of log file sync. Log file sync it can be essential more write time on a disk because of the same group commit.
If there are problems it is necessary to reduce kol-in commit. If for any reason it is not pleasant AWR anything it is not necessary to do.

13

Re: High waitings log file sync

Long (on the average) is not present
Simply short it is too much ( if to estimate that the standard report becomes in an hour, ~8000 waitings in a second explicit search)

14

Re: High waitings log file sync

Ljubomudrov wrote:

Long (on the average) is not present
Simply short it is too much ( if to estimate that the standard report becomes in an hour ~8000 waitings in a second explicit search)

Well if is hour at it 60 kernels on 100 % are loaded, and I doubt it.

15

Re: High waitings log file sync

Well in general, kol-in waitings depends only from kol-va . If there is a real problem, for example when 3ms this average, but is  it is necessary to look the histogram on log file parallel write from this the report and a broad gull file LGWR where time delays separate  go.
And if it simply attempt on the basis of TOP 5 from AWR to start to be engaged in tuning it is unpromising the approach.