| log switches [message #51679] |
Thu, 06 June 2002 01:44  |
anthony smith
Messages: 1
Registered: June 2002
|
Junior Member |
|
|
Hi all,
I wanted some help with interpreting exactly what is going on with my log switches below. These are 4x 1GB redo logs.
EXHIBIT A
#### Here we have overlapping checkpoints
##############################################
Mon Jun 3 09:54:39 2002
Beginning log switch checkpoint up to RBA [[0x1707.2.10]], SCN: 0x0358.972df5a5
Thread 1 advanced to log sequence 5895
Mon Jun 3 10:01:40 2002
Beginning log switch checkpoint up to RBA [[0x1708.2.10]], SCN: 0x0358.972e0527
Thread 1 advanced to log sequence 5896
Mon Jun 3 10:01:41 2002
Completed checkpoint up to RBA [[0x1707.2.10]], SCN: 0x0358.972df5a5
Mon Jun 3 10:02:05 2002
Completed checkpoint up to RBA [[0x1708.2.10]], SCN: 0x0358.972e0527
Mon Jun 3 10:23:12 2002
Mon Jun 3 10:23:12 2002
Beginning log switch checkpoint up to RBA [[0x1709.2.10]], SCN: 0x0358.972e3b2b
Thread 1 advanced to log sequence 5897
Mon Jun 3 10:33:48 2002
Completed checkpoint up to RBA [[0x1709.2.10]], SCN: 0x0358.972e3b2b
EXHIBIT B
## Here we have checkpoints taking around 6 mins (and around 5mins to complete) and only then does another occur!!!!
####################################################################################################################
Mon Jun 3 12:20:49 2002
Beginning log switch checkpoint up to RBA [[0x1714.2.10]], SCN: 0x0358.97fb5ff3
Thread 1 advanced to log sequence 5908
Mon Jun 3 12:25:34 2002
Completed checkpoint up to RBA [[0x1714.2.10]], SCN: 0x0358.97fb5ff3
Mon Jun 3 12:26:14 2002
Beginning log switch checkpoint up to RBA [[0x1715.2.10]], SCN: 0x0358.97fb8ad5
Thread 1 advanced to log sequence 5909
Mon Jun 3 12:30:41 2002
Completed checkpoint up to RBA [[0x1715.2.10]], SCN: 0x0358.97fb8ad5
Mon Jun 3 12:31:20 2002
Beginning log switch checkpoint up to RBA [[0x1716.2.10]], SCN: 0x0358.97fbb5af
Thread 1 advanced to log sequence 5910
Mon Jun 3 12:35:38 2002
Completed checkpoint up to RBA [[0x1716.2.10]], SCN: 0x0358.97fbb5af
Mon Jun 3 12:36:14 2002
Beginning log switch checkpoint up to RBA [[0x1717.2.10]], SCN: 0x0358.97fbe0e9
Thread 1 advanced to log sequence 5911
Mon Jun 3 12:43:00 2002
Completed checkpoint up to RBA [[0x1717.2.10]], SCN: 0x0358.97fbe0e9
Mon Jun 3 12:43:30 2002
Beginning log switch checkpoint up to RBA [[0x1718.2.10]], SCN: 0x0358.97fc04f1
Thread 1 advanced to log sequence 5912
Mon Jun 3 12:48:03 2002
Completed checkpoint up to RBA [[0x1718.2.10]], SCN: 0x0358.97fc04f1
INIT.ORA params
db_block_size = 16384
db_block_buffers = 135000
dbwr_io_slaves = 4
db_writer_processes = 1
log_buffer = 1048576
log_checkpoint_interval = 9999999999
log_checkpoint_timeout = 0
disk_asynch_io = TRUE
I have read that during a log switch a lock is taken on the datafiles
as they are updated with new SCN etc and that no redo can be generated during this time.
1)
If this is the case then is my database inactive during the MINUTES it takes to complete a log switch?
(Exhibit A suggest not but exhibit B suggests so.)
if so what can I do about it. I have read a lot on this list and elsewhere on this but feel that I have done everything suggested so far.
If it is a DBWR issue then I could look at the amount DBWR writes, using more than one DBWR process and so on but a solution of increasing the redo log size does
not seem ideal to be given that they are alread 1GB!
rgds
anthony
|
|
|
|
| Re: log switches [message #51681 is a reply to message #51679] |
Thu, 06 June 2002 05:22  |
Alex Shenfield
Messages: 15
Registered: May 2002
|
Junior Member |
|
|
Hi,
You need to understand why so much redo is being written. Check the size of the archive logs are they 1GB or smaller? You need to understand whats happening at the application level. This could be a dataload in which case you might not need to write the archive?
Check points & log switchs are very costly but don't completely stop the database.
Ideally for performance reasons you should have you redo log file on there own disks/controllers. The groups members on differant disks/controllers thu other groups can be on the same disk/controllers
|
|
|
|
] 
Blog
Search
Help
Members
Register
Login
Home
