why is this happenning occasionally I can see some bg processes like snp0,snp1 are dying ,sometimes shared server ,dispatchers getting dead.what 's happening??i can see the errors in alertlog.
pl reply back,its urgent
Printable View
why is this happenning occasionally I can see some bg processes like snp0,snp1 are dying ,sometimes shared server ,dispatchers getting dead.what 's happening??i can see the errors in alertlog.
pl reply back,its urgent
Can you post the relevant error numbers and messages from the error log? Also, are any trace files written? Are there any patterns you are seeing like it is happening during a load, during the execution of a certain procedure, etc..
Joe
found dead multi-threaded server 'S000', pid = (10, 1)
Tue Nov 28 08:30:59 2000
DBW0: terminating instance due to error 472
Instance terminated by DBW0, pid = 165
this is the message of alert log.No i cannot see any perticular procedure causing that.how can i trace that any procedure is causing it ?there r so many users.most of the times
isee-
restarting dead background process snp0
this is the last statement in alert log,after this statement i can see i cannot connect to sql*plus.so most often snp are not getting started.do u want anything more??
If you have configured MTS, then increase maximum number of servers, and dispatches in INIT.ORA file.
Monitor the background processes using a shell script.
Is your SGA size less than available memory?
hi tamil,
the no. of server proceses is 5,as no. of users are 24.dispatchers-3,do u think it is needed to increase this number.i am on NT,so how to trace actual cause of errors.
sga size 370mb and ram 512 mb.what's the cause??pl reply back ,otherwise I will have to reinstall oracle.as this problem is causing lots of work to stop.how to monitor BG proceses on NT.
What Version of Oracle, Patchsets, NT Service Packs?
Update:
I found quite a lot of hits on Metalink associated with this. There seems to be some known bugs on certain versions.
On common piece of advice is to remove Dead Connection Detection from the sqlnet.ora file on the machine on which the instance is located:
i.e. set:
sqlnet.expire_time=0
Are you getting any ORA-600 messages produced?
[Edited by TerryD on 11-29-2000 at 04:45 AM]
hi everybody,
I increased the no.of servers to 7 as tamil said but constantly this error message is coming in alertlog-
skgpspawn failed:category = 27143, depinfo = 9261, op = spdcr, loc = skgpspawn
failed to start multi-threaded server 'S006', oer=27143
whats the problem???
its oracle 8.1.5 on NT 4.0 service pack 5.ora-600 I also saw in alertlog some days before,how to get rid of it?
any patches??
another thing -if at server i connect by giving scott/tiger I see ora-12560 tns protocol adapter error.
but if I issue scott/tiger@orcl its getting connected.this is happenning occasionally,if i restart for sometime the problem gets solved ,but again after sometime it comes.
thanks for all help.any suggestion is most welcome.
Well, I found this on a link about a very similar problem:
The ORA-600:[729] error is likely caused by a bug which effects 8.1.5 where a shared server process can encounter an ORA-600:[729] error when a process using database links disconnects. The bug is fixed in 8.1.6 and included in the 8.1.5.1 patchset release. The 8.1.5.1.1 patchset release for NT is available for download from the Patches option on MetaLink Home.
I've also noticed a couple of other links relating to SP5.
The common suggestion though is to set:
sqlnet.expire_time=0
in the sqlnet.ora file on the server, and see if that helps. The other suggestion is to raise a TAR with Oracle - I think they will come up with quite a few hits relating to this. Although you will need the whole ORA-600 message and any associated trace files.
let me tell u this is workgroup server ,so pl tell me whats the limits of workgroup server,can i use any patch to overcome this error.
but first tell whats diff betwn workgroup and enterprise edition,i know partitioning,transportable tbsp,index organized tables are not supported here.except these features if there is any limits set .
some other errors in alertlog-
Errors in file E:\ORACLE\admin\ORTELAPP\bdump\ortelappSNP1.TRC:
ORA-00604: error occurred at recursive SQL level 1
ORA-04030: out of process memory when trying to allocate 20 bytes (callheap,heap mark)
skgpspawn failed:category = 27143, depinfo = 9261, op = spdcr, loc = skgpspawn
failed to start dispatcher 'D002' for network '(ADDRESS=(PARTIAL=YES)(PROTOCOL=TCP))', oer=27143
whats that memory error.
Shortage of memory is your real problem.
You should decrease the SGA size to 30-40% of RAM.
hi tamil,
can u tell why the sga size should be in that ratio to my RAM.
I read in some books that I can give more to SGA and little memory is needed for NT BG processes,I will really appreciate any help.
this is what i read-
for optimal performance,u want to make sure that u'r machine has enough memory to accomodate the entire sga and still have some memory left for OS to perform well.pl reply back asap.it will be a great help.
Would it be a good idea to write some FAQ's up or publish some already created faq's on the site? I see a large number of repeated common errors and it seems it would save a lot of postings and such to make them available.
Joe
Could you please email me the following:
1. A copy of your initialization file
2. A copy of your tnsnames and listener files
3. A copy of your alert log
4. Include in the text of your email #CPU's and memory you
have.
5. At a time when you have an average load (watch out for
lunch breaks) run the following:
SVRMGRL> your_oracle_home\rdbms\admin\utlbstat
This will run some scripts to create some temporary tables and collect some statistics.
Let this run for about 2 hours and then run
SVRMGRL> your_oracle_home\rdbms\admin\utlestat
Make sure before you do this that the parameter TIMED_STATISTICS = TRUE. Also, when you run utlestat, make sure you are in a directory that you have write permissions to.
There will be a report generated called report.txt - attach this file as well to your email and send it my way.
Thanks,
Joe
hi wjramsay,
I donot have your mail address ,pl give that.as soon as possible I will send all that U needed.as per U asked about creating FAQs will definitely be helpful.it will be a help also
to give the list of bugs each version of oracle have and the corresponding patches available by metalink.thanks for spending so much of time for my queries.
regards
smruti
hi,
only the snapshot processes r getting dead occasionally.and wjramsay pl let me have u'r email address so that I can give those files as u asked.we donot have oracle support so I cannot avail patches from metalink.any other reply is welcome
and tamil
if u get some time pl let me know about that sga size to be 30-40% of RAM.
warm regards and personally I thank u all for all u'r helps
I believe it's in my profile but you can email the items
to
[email protected]
Thanks :)
Joe
HI JOE,
LASTLY i REDUCED THE SIZE OF SHARED_POOL WHICH WAS SET TO 230M TO 200M AND DB_BLOCK_BUFFERS TO 50000 WHICH WAS 60000 AND WATCHING FOR FURTHER HAPPENINGS,
IF THE PROBLEM STILL CONTINUES THEN i WILL LET U KNOW
AT U'R MAIL ADDRESS.
REGARDS AND THANKS
I AM READING IN THIS FORUM FROM LONG TIME TO SET SGA TO 30-40% OF MEMORY.BUT IN SOME BOOKS i FOUND SOMETHING ELSE,LIKE IF NO OTHER APPLICATION RUNNING IN SERVER I CAN SET SGA TO MORE THAN 50%.
WHAT DO U SAY?IN PREV VERSIONS OF ORACLE SOME VIEWS LIKE CURRENT_BUCKET AND RECENT_BUCKET WAS THERE TO FIND OUT THE NO. OF BLOCK BUFFERS NEEDED TO MAKE THE BUFFER HIT RATION >90%.IN VER 8.1.5 THEY R NOT THERE?HAVE U GOT ANY OTHER WAY TO FIND THE REQUIRED NO.OF BLOCK BUFFERS TO INCREASE HIT RATIO.OR JUST TO INCREASE AND CHECK THE RESULT.