Hi,
We intend to migrate to RAC shortly (11gR2).
On the current single-instance databases we have the following:
SQL> select inst_id,group#,thread# ,bytes from gv$log order by 1,2,3;
INST_ID GROUP# THREAD# BYTES
---------- ---------- ---------- ----------
1 1 1 3221225472
1 2 1 3221225472
1 3 1 3221225472
1 4 1 3221225472
(each group has as single file on a different filesystem every time, therefore 4 filesystems in total).
In a RAC environment, how the select results above should look like ? Should I see 8 lines with
4 groups for each instance ID or rather 16 lines... ? Viz should we get
SQL> select inst_id,group#,thread# ,bytes from gv$log order by 1,2,3;
INST_ID GROUP# THREAD# BYTES
---------- ---------- ---------- ----------
1 1 1 3221225472
1 2 1 3221225472
1 3 2 3221225472
1 4 2 3221225472
2 1 1 3221225472
2 2 1 3221225472
2 3 2 3221225472
2 4 2 3221225472
or rather:
SQL> select inst_id,group#,thread# ,bytes from gv$log order by 1,2,3;
INST_ID GROUP# THREAD# BYTES
---------- ---------- ---------- ----------
1 1 1 3221225472
1 2 2 3221225472
1 3 1 3221225472
1 4 2 3221225472
1 5 1 3221225472
1 6 2 3221225472
1 7 1 3221225472
1 8 2 3221225472
2 1 1 3221225472
2 2 2 3221225472
2 3 1 3221225472
2 4 2 3221225472
2 5 1 3221225472
2 6 2 3221225472
2 7 1 3221225472
2 8 2 3221225472
And does it mater how the thread number alternate with the groups ? (pay attention above to
the sequence in the thread numbers along with the group numbers...) Should I really worry
about that ? (an info that might be important: our groups in RAC will also contain only one
member -my SAN admin told me it wouldn't have sense *at all* to multiplex the members in
this future environment...-)
The answer might seem obvious to some of you, but frankly from the doc. it does not appear clear-cut
to me how it should be configured...
Thanks a lot for any enlightenment !
Regards,
Seb