select username, machine, osuser, to_char(logon_time, 'YYYY-MM-DD HH24:MI') LG ,last_call_et from gv$session where username = 'GDBA_WUQIH' GDBA_WUQIH ASIAPACIFIC\SCN7872 wuqih 2007-09-27 17:23 0 GDBA_WUQIH ASIAPACIFIC\SCN7872 wuqih 2007-09-27 16:40 0 GDBA_WUQIH ASIAPACIFIC\SCN7872 wuqih 2007-09-27 16:40 0 GDBA_WUQIH ASIAPACIFIC\SCN7872 wuqih 2007-09-27 17:23 0 From the result, the session connects to two instances at the same time. it's amazing. Is it correct that a session connect to two nodes? On 9/27/07, LS Cheng <exriscer@xxxxxxxxx > wrote: > > I dont work with PL/SQL Developer so I am not sure what it does underwood. > But this is very easy to check > > select username, machine, module, osuser, to_char(logon_time, 'YYYY-MM-DD > HH24:MI') LG ,last_call_et > from gv$session > where username = 'GDBA_WUQIH' > > Thanks > > -- > LSC > > > > On 9/27/07, qihua wu <staywithpin@xxxxxxxxx > wrote: > > > > select inst_id,count(*) from gv$session where schemaname='GDBA_WUQIH' > > group by inst_id > > 1 2 > > 2 2 > > > > And I have only two session through plsql devloper. Each of these > > sessions lands on both instances. That surprise me. One session sits on both > > instance at the same time. > > > > On 9/27/07, LS Cheng <exriscer@xxxxxxxxx> wrote: > > > > > > please add a predicate for your user and not groupping at global level > > > because for a idle RAC system there are quite a few background processes > > > and > > > RAC monitoring processes and if you have dbconsole then more > > > > > > thanks > > > > > > -- > > > LSC > > > > > > > > > > > > On 9/27/07, qihua wu <staywithpin@xxxxxxxxx > wrote: > > > > > > > > I don't know why ask for groupping by inst_id, anyway the result is. > > > > select inst_id,count(*) from gv$session group by inst_id > > > > 1 36 > > > > 2 39 > > > > > > > > David may be correct that sid is not unique accross instance, but > > > > why gv$mystat will show non-zero stats value on both instances(since > > > > non-zeror statatics, that means the session operates on both instance). > > > > > > > > Dan: > > > > a session connects to the database multiple times, it still has the > > > > same session id, otherwise they are not the same session, right? > > > > > > > > My question is "one session can only land on one instance during its > > > > lifetime?". > > > > > > > > Thanks very much for your reply. Have a good day. > > > > > > > > On 9/26/07, LS Cheng <exriscer@xxxxxxxxx> wrote: > > > > > > > > > > how about group by inst_id...? > > > > > > > > > > thanks > > > > > > > > > > -- > > > > > LSC > > > > > > > > > > > > > > > On 9/27/07, qihua wu < staywithpin@xxxxxxxxx > wrote: > > > > > > > > > > > > We have a two nodes RAC, the strange thing is that many sessions > > > > > > run on both nodes. > > > > > > select sid from gv$session group by sid having count(*)>1 > > > > > > they are lots of rows return. > > > > > > > > > > > > My understanding for load balance is: if sessions on instance 1 > > > > > > are too many, then new session will be connected to instance 2. > > > > > > But if at the beginning there were100 sessions on both inst1 and > > > > > > inst2, after all the sessions on inst2 are completed, no existing > > > > > > session on > > > > > > instance1 will go to instance2. > > > > > > > > > > > > But the fact of the above query told me some sessions run on > > > > > > both instance. How can this be true? > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >