Path: news.cambrium.nl!textnews.cambrium.nl!feeder1.cambriumusenet.nl!feed.tweaknews.nl!193.201.147.78.MISMATCH!feeder.news-service.com!postnews.google.com!o6g2000yqj.googlegroups.com!not-for-mail From: zigzagdna Newsgroups: comp.databases.oracle.server Subject: Re: Why do I get out of memory errors when 10GB memory is free? Date: Tue, 7 Jul 2009 08:43:44 -0700 (PDT) Organization: http://groups.google.com Lines: 113 Message-ID: References: <405e0eb3-5ced-41fb-9769-e570695f8a03@g19g2000yql.googlegroups.com> <7371c37a-de6c-41e9-8380-129dc3ad09d5@t33g2000yqe.googlegroups.com> NNTP-Posting-Host: 148.177.0.100 Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: quoted-printable X-Trace: posting.google.com 1246981425 19830 127.0.0.1 (7 Jul 2009 15:43:45 GMT) X-Complaints-To: groups-abuse@google.com NNTP-Posting-Date: Tue, 7 Jul 2009 15:43:45 +0000 (UTC) Complaints-To: groups-abuse@google.com Injection-Info: o6g2000yqj.googlegroups.com; posting-host=148.177.0.100; posting-account=PUA9dQoAAAAWB6fqjaJDJmTx7eb4x-f4 User-Agent: G2/1.0 X-HTTP-UserAgent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022; InfoPath.2; .NET CLR 3.0.4506.2152; .NET CLR 3.5.30729; .NET CLR 1.1.4322; MS-RTC LM 8),gzip(gfe),gzip(gfe) Xref: news.cambrium.nl On Jul 7, 11:26=A0am, zigzagdna wrote: > On Jul 6, 6:11=A0pm, joel garry wrote: > > > > > > > On Jul 4, 6:32=A0pm, zigzagdna wrote: > > > > On Jul 4, 9:15=A0pm, zigzagdna wrote: > > > > > 1. =A0 =A0 =A0gpsgetpd=3D>ulimit -a > > > > time(seconds) =A0 =A0 =A0 =A0unlimited > > > > file(blocks) =A0 =A0 =A0 =A0 unlimited > > > > data(kbytes) =A0 =A0 =A0 =A0 1048576 > > > > stack(kbytes) =A0 =A0 =A0 =A0131072 > > > > memory(kbytes) =A0 =A0 =A0 unlimited > > > > oredump(blocks) > > > > > 2. =A0 =A0 =A0Some kernel parameters relevant to oracle: > > > > shmmax =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A025662429593 =A025= 662429593 =A0 =A0 =A0 =A0 Immed > > > > shmmni =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A05= 12 =A0512 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 Immed > > > > shmseg =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A01= 20 =A0120 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 Immed > > > > max_thread_proc =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A01024 =A01024= =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0Immed > > > > maxdsiz =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A01073741824 =A0De= fault =A0 =A0 =A0 =A0 =A0 =A0 Immed > > > > maxdsiz_64bit =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A02147483648 =A021474836= 48 =A0 =A0 =A0 =A0 =A0Immed > > > > maxfiles =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 5120 = =A05120 > > > > maxfiles_lim =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 6144 =A061= 44 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0Immed > > > > maxrsessiz =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A08388608 =A0De= fault > > > > maxrsessiz_64bit =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A08388608 =A0Default > > > > maxssiz =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 134217728 =A013= 4217728 =A0 =A0 =A0 =A0 =A0 Immed > > > > maxssiz_64bit =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A01073741824 =A010737418= 24 =A0 =A0 =A0 =A0 =A0Immed > > > > maxtsiz =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 100663296 =A0De= fault =A0 =A0 =A0 =A0 =A0 =A0 Immed > > > > > I do not see anything wrong (low values) for ulimit or anyting wron= g > > > > in kernel parameters. > > > > Here are some observations related to memory: > > > 1. =A0 =A0 =A0I can increase sga_target (that=92s the only parameter = I have for > > > SGA) by several =A0Giga Bytes without any problem, i.e., Oracle insta= nce > > > starts with any memory issues. > > > 2. =A0 =A0 =A0If I start a java process, by simply running java comma= nd, it gives > > > me out of memory error . It seems that sever does not have enough > > > emmory for heap (evne though 10GB of memory is free on server) becaus= e > > > java needs some default heap memory (I think 64M).. IT cannot be a > > > ulimit issue because maxdsiz is 4 GB. > > > 3. =A0 =A0 =A0If I reduce sga_target to a smaller value, restart orac= le instance > > > and then run java, java does not give memory error, so Oracle freed > > > some memory. What I do not understand is Oracle is freeing shared > > > memory and not heap memory, so why it has impact on java (i.e., java > > > getting more heap memory). Server all anog has at least 9 to 10GB > > > memory free. > > > 4. =A0 =A0 =A0Problem does not happen with Oracle startup, oracle is = running > > > fine. Oracle export command and rman =A0off and on fail with ORA erro= r > > > related to memory. > > > 5. =A0 =A0 =A0My server has 24GB memory and in all above cases, at le= ast 40% of > > > 24GB is free. > > > > Hope someone can point what can be done to get rid of memory errors. > > > What is JAVA_POOL_SIZE SHARED_POOL_SIZE =A0LARGE_POOL_SIZE > > > What exact version of java are you using? =A0When you run java, the hea= p > > size that is asked for can actually determine that it will run > > different java programs. > > > Your maxssize may be too small (as Mladen hinted), seehttp://docs.hp.co= m/en/JAVAPERFTUNE/Memory-Management.pdfunderkernel > > tunables. > > > It doesn't really matter if a lot of memory is free if you've limited > > the size of what is asked for. > > > jg > > -- > > @home.com is bogus.http://www3.signonsandiego.com/stories/2009/jul/06/u= s-tec-microsoft-s...Hide quoted text - > > > - Show quoted text - > > I talked to HP engineer. Per HP engineer there is some kind of > limtation in HP arhicteure that for 32 bit processes, it needs 1GB > contigous =A0space (even if 32GB process is not using 1GB memory). On > that server even though there is lot of free memory, there is not > contiguous 1Gb free memory. So root cause of the problem is something > different than kernel parameters.- Hide quoted text - > > - Show quoted text - I am still talking to HP engineer because rman backup which fails off and on is a 64 bit executable not 32 bit executable.