关于checkpoint

wing hong

最初由 husthxd 发布
[B]意犹未尽.
顶上去.继续......

那位xd来解析一下tom下面的这段话?
the infamous "checkpoint not complete, cannot allocate new log" message.

this occurrs when Oracle attempts to reuse a log file but the checkpoint that would flush the blocks that may have redo in this log file has not yet completed
-- we must wait until that checkpoint completes before we can reuse that file -- thats when this message is printed.  during this time (when we cannot allocate a new log) processing is suspended in the database while the checkpoint is made to complete ASAP. [/B]

you need to understand this from the point of crash recovery.

During the time of crash recovery, Oracle will do the recovery automatically. so all Oracle has is the online redo log file.

So if Oracle allows the checkpoint not finish but reuse the redo log , then if Oracle need to do crash recovery, it must need the archived log file.  then the crash recovery is not automatic because the archived log file can be moved .

wing hong

because Oracle's recovery process is roll forward then roll back. so if Oracle need to do a recovery at RBA point A, all the data blocks that are used in the redo record  before RBA point A must be on the disk. then Oracle can start to rollforward by applying the redo log.

Instance/Crash recovery is similar except it only has the online redo log to use.

husthxd

got it,3k

sydongsun

关于checkpoint和恢复过程的详细全面的资料, 也可以参考了:
http://www.askguoyu.com/db/detailofArchtecture.htm

realji

不是很明白

roman2000

切换日志会触发checkpoint
什么是checkpoint?
    checkpoint是一个数据库事件，它将已修改的数据从高速缓存刷新到磁盘，并更新控制文件和数据文件。
   
    什么时候发生checkpoint?
    我们知道了checkpoint会刷新脏数据，但什么时候会发生checkpoint呢？以下几种情况会触发checkpoint。
    1.当发生日志组切换的时候
    2.当符合LOG_CHECKPOINT_TIMEOUT，LOG_CHECKPOINT_INTERVAL，fast_start_io_target,fast_start_mttr_target参数设置的时候
    3.当运行ALTER SYSTEM SWITCH LOGFILE的时候
    4.当运行ALTER SYSTEM CHECKPOINT的时候
    5.当运行alter tablespace XXX begin backup，end backup的时候
    6.当运行alter tablespace ,datafile offline的时候;

roman2000

贴错地方了,不好意思

wxlmail2008

引自roman2000
切换日志会触发checkpoint
什么是checkpoint?
checkpoint是一个数据库事件，它将已修改的数据从高速缓存刷新到磁盘，并更新控制文件和数据文件。

什么时候发生checkpoint?
我们知道了checkpoint会刷新脏数据，但什么时候会发生checkpoint呢？以下几种情况会触发checkpoint。
1.当发生日志组切换的时候
2.当符合LOG_CHECKPOINT_TIMEOUT，LOG_CHECKPOINT_INTERVAL，fast_start_io_target,fast_start_mttr_target参数设置的时候
3.当运行ALTER SYSTEM SWITCH LOGFILE的时候
4.当运行ALTER SYSTEM CHECKPOINT的时候
5.当运行alter tablespace XXX begin backup，end backup的时候
6.当运行alter tablespace ,datafile offline的时候;

谢谢

hanjs

原帖由 sydongsun 于 2005-4-17 22:32 发表
关于checkpoint和恢复过程的详细全面的资料, 也可以参考了:
http://www.askguoyu.com/db/detailofArchtecture.htm

网页都不存在了！

有点晕，再来一遍

foxmile

原帖由 wing hong 于 2005-3-31 20:10 发表
because Oracle's recovery process is roll forward then roll back. so if Oracle need to do a recovery at RBA point A, all the data blocks that are used in the redo record  before RBA point A must be on the disk. then Oracle can start to rollforward by applying the redo log.

Instance/Crash recovery is similar except it only has the online redo log to use.

这个很经典。一语道破两个恢复的玄机。
如果checkpoint一直不complete，有什么后果。