You are right (ofcourse). Did not know that. But found it in the
documentation (Concepts Guide).
Howard J. Rogers <dba_at_hjrdba.com> schreef in berichtnieuws
afih7g$oqv$1_at_lust.ihug.co.nz...
|
| "Anton Buijs" <aammbuijs_at_xs4all.nl> wrote in message
| news:afi2i0$g18$1_at_news1.xs4all.nl...
| > The answer is in the values of all the log_arch% parameters from the
| > init.ora file or current values from V$PARAMETERS
| >
|
| The true answer is that Oracle will dynamically spawn additional archiver
| processes when it notices that a backlog of online redo logs for archiving
| is building up. It's trying to do you a favour, and prevent the very nasty
| system stall you'll get when you try and switch back to an online log
which
| has been checkpointed, but which hasn't yet been archived.
|
| For example, You switch from log 1 to log 2, and then quickly to log 3 and
| 4. You've only got log 5 between you and switching back to log 1.
Meanwhile,
| the solo archiving process is still plodding along, archiving Log 1.
Oracle
| will notice that 2, 3 and 4 also need to be archived, so spawns additional
| processes to get them archived. When you finally switch into log 1, you'd
| discover that logs 2, 3 and 4 have also been archived. There's thus a
clear
| run between you and log 5.
|
| Oracle is then supposed to kill off the additional archiver process, and
| re-spawn them if needed. But I've rarely seen it do it. The additional
| processes seem to hang around for ever, idle. Not sure if that's a (minor)
| bug, or a feature, or operating system dependent.
|
| It is also true that you can set a log_archive_max_processes init.ora
| parameter. This does not do what it says it will do. It *sounds* as though
| this will place a *limit* on the number of archivers Oracle is allowed to
| spawn dynamically. In fact, it indicates the number of archivers you want
| created at instance startup. Oracle will quite happily spawn processes in
| excess of this "maximum" if it feels the need.
|
| Regards
| HJR
|
| Regards
| HJR
|
|
|
|
|
|
| > Vince Laurent <vlaurent_at_NOSPAM.networkusa.net> schreef in berichtnieuws
| > 8i1phu4m7uu52a7uoh7j5cvhjitvbps10c_at_4ax.com...
| > | On one of our systems I see the following behavior:
| > |
| > | (sapbpr):/oracle/BPR/saptrace/background:root# ps -ef | grep -v grep |
| > | grep ora_
| > | bpradm 18215 1 0 Jun 19 ? 0:42 ora_dbw0_BPR
| > | bpradm 18217 1 0 Jun 19 ? 1:27 ora_lgwr_BPR
| > | bpradm 18213 1 0 Jun 19 ? 0:15 ora_pmon_BPR
| > | bpradm 18225 1 0 Jun 19 ? 0:35 ora_arc0_BPR
| > | bpradm 18221 1 0 Jun 19 ? 0:04 ora_smon_BPR
| > | bpradm 18223 1 0 Jun 19 ? 0:00 ora_reco_BPR
| > | bpradm 18219 1 0 Jun 19 ? 0:47 ora_ckpt_BPR
| > | bpradm 4242 1 0 14:43:11 ? 0:01 ora_arc1_BPR
| > |
| > | And from the alert log:
| > |
| > | Completed checkpoint up to RBA [0xfa.2.10], SCN: 0x0000.0053776d
| > | Thu Jun 20 14:43:04 2002
| > | Beginning log switch checkpoint up to RBA [0xfc.2.10], SCN:
| > | 0x0000.0053802b
| > | Thread 1 advanced to log sequence 252
| > | Current log# 14 seq# 252 mem# 0: /oracle/BPR/origlogB/log_g14m1.dbf
| > | Current log# 14 seq# 252 mem# 1: /oracle/BPR/mirrlogB/log_g14m2.dbf
| > | Thu Jun 20 14:43:11 2002
| > | ARC0: Completed archiving log# 11 seq# 250
| > | ARC0: STARTING ARCH PROCESSES
| > | ARC0: STARTING ARCH PROCESSES COMPLETE
| > | ARC0: Beginning to archive log# 12 seq# 251
| > | ARC1 started with pid=34
| > | ARC1: Archival started
| > | Thu Jun 20 14:43:21 2002
| > | ARC0: Completed archiving log# 12 seq# 251
| > |
| > | So why does Oracle start a second archive process?
| > |
| > | We are running Oracle 8.1.7.3 on HP-UX 11.x
| > |
| > | Thanks!
| > |
| >
|
|
Received on Fri Jun 28 2002 - 15:52:53 CDT
