Re: Why event "global cache lock open x " wait forever?

Greetings,

I suggest get the p3 value from v$session_wait , it is the lock element, and try to query gv$session_wait , include the inst_id in your query and order by
p3. If the same lenum is being used by other instances, then it may be a bug.

since you already know the table and index, move them to its own tablespace
and allocate sufficent gc_files_to_locks. Hope yu have freelists and freelist groups allocated. It is very important. If that doesn't help try investigating
about adding the following parameters to your inidSID.ora file on all the OPS nodes.

_upconvert_from_ast       = FALSE
_ignore_failed_escalates  = FALSE
_save_escalates           = FALSE

If it still doesn't help, use the delete_commit procedure to delete a specific number of rows as suggested by Mark Powell.

Hope This Helps.
Regards & thanks
BN

sroesch3_at_attbi.com (Steve) wrote in message news:<efc3f20b.0210101717.6c09aa29_at_posting.google.com>...
> Advice so far doesn't make a lot sense for me.
>
> Version 8.05 isn't supported now for a very long time. Think about
> upgrading to a supported release.
>
> First thing is to figure out if it is slow or actually hanging.
> To check that enable event 10046 level 8 and look in the trace
> file if it is stuck or simply slow.
>
> The wait event global cache open x means that we are trying to
> open a new lock in exclusive mode. Unfortunately you didn't
> mention p1 and p2 which is the fileno and the block no.
>
> Next step is to figure out which kind of class the object belongs
> to and how many locks/resources are configured for that file. This also
> depends on what kind of locking you are using (hash or dba locking)
>
> You might check the parameter gc_file_to_locks
>
>
> Regards
> Steve
>
>
>
>
> melocy_at_21cn.com (melocy) wrote in message news:<92A2664ABmelocy21cncom_at_130.133.1.4>...
> > Evironment:2 * RS6000 AIX4.3.2 Oracle805 OPS(db1,db2)
> > on db2,one transaction is that:
> >
> > set transaction use rollback segment rbig;
> > delete from aaa.t1;
> > insert into aaa.t1 select * from bbb.t1;
> > commit;
> >
> > WHEN execute statement "Insert into ....",
> > process hang.
> > query v$session_wait,event is "global cache lock open x "
> > query v$session_event,event "global cache lock open x"
> > total_waits is adding continually and total_timeouts is adding too.
> >
> > Help me ,Pls.
> > Thx.
Received on Fri Oct 11 2002 - 08:43:36 CDT