database crash ora-1575

Hunan_1 — Sat, 07 Dec 2002 02:57:53 GMT

hi all,
please I'm in problems, I work with an older oracle database (release 7.1.6.2).
suddenly in the alert log is being inserted the ora-1575 error very often, this message was shown before but not too often, in consecuence the application using the database hangs.
I've made some queries that look for contention, but it didn't show anything.

Any help would be really appreciated

Re: database crash ora-1575

Yogeeraj_1 — Sat, 07 Dec 2002 05:23:11 GMT

Hi,

Below a quote from Metalink:

Problem description:
=====================
You are experiencing a performance degredation and/or are seeing the following errors reported in your ALERT.LOG:

ORA-01575: timeout waiting for space management resource Cause: failed to acquire necessary resource to do space management.
Action: Retry the operation.

The INIT.ORA parameter REDUCE_ALARM is set to TRUE.

Problem Explanation:
====================
In 7.0.12 the optional INIT.ORA parameter REDUCE_ALARM which enables the Wakeup Montior (WMON) process was introduced to alleviate contention on the UNIX callout queue by shifting workload to the SGA. The appearance of occasional ORA-01575 errors in your ALERT.LOG are a normal side-effect of this WMON process running. However, it is possible that the WMON process may cause performance degradation by increasing contention for SGA latches.

Solution Description:
=====================
The appearance of ORA-01575 errors is a side-effect of enabling the WMON process via the INIT.ORA parameter:

reduce_alarm=TRUE

Oracle suggests that the reduce_alarm parameter not be used, or be set to FALSE, unless a measurable performance gain is perceived. In general, enabling WMON is only useful in UNIX based OLTP environments experiencing a high volume of commit operations.

Moreover, in instances where large numbers of latch waits exist, enabling of WMON may actually result in a performance deterioration because WMON shifts workload from the UNIX callout queue to areas in the SGA resulting in increased latch contention in the SGA.

If your are not in an OLTP environment, reduce_alarm is not recommended because it results in an on-beneficial increase in SGA activity.

Solution Explanation:
=====================
The WMON process was introduced to alleviate a bottleneck on the UNIX callout queue during LGWR posting by shifting workload from the callout queue to an area in the SGA. However, if your system already has a high number of latch waits, the WMON process may exacerbate contention problems for latches in the SGA resulting in a performance degradation of your RDBMS. As such, Oracle only recommends setting this paramter to TRUE when a performace gain is perceived. Generally speaking, OLTP environments with a high volume amount of commits would be the only instances where WMON could prove benefical.

The ORA-01575 errors associated with enabling WMON are a result of increased SGA enqueue timeouts by SMON. In essence, an ORA-1575 error only indicates that the SMON cannot coalesce. Occasional ORA-1575's do not indicate trouble and are to be expected when the load on the SGA is high since SMON uses timeouts to prevent it from waiting indefinately on an enqueue when it has other tasks to handle such as preventing deadlocks.

Hope this helps!

Best Regards
Yogeeraj

Re: database crash ora-1575

Hunan_1 — Sun, 08 Dec 2002 19:58:39 GMT

I'll apply your recomendations tomorrow, thanks for your helping

topic Re: database crash ora-1575 in Operating System - HP-UX

database crash ora-1575

Re: database crash ora-1575

Re: database crash ora-1575

Re: database crash ora-1575