Problem solved: The cause of the problem is another Oracle installation on this server, specifically an Oracle Restart. I used sqlplus to shutdown immediate the database instance, but 23ai still failed to install. After "crsctl stop has", I was able to run “/etc/init.d/oracle-free-23ai configure”, and the 23ai FREE instance is up and running.
Original message follows:
I'm trying to install 23ai with an RPM on RHEL9.5. Here's what I did:
# wget https://yum.oracle.com/RPM-GPG-KEY-oracle-ol9 -O /etc/pki/rpm-gpg/RPM-GPG-KEY-oracle
# gpg --quiet --keyid-format 0xlong --with-fingerprint /etc/pki/rpm-gpg/RPM-GPG-KEY-oracle
The above is from https://yum.oracle.com/faq.html#a10
# vi /etc/yum.repos.d/ol9-temp.repo
[ol9_baseos_latest]
name=Oracle Linux 9 BaseOS Latest ($basearch)
baseurl=https://yum.oracle.com/repo/OracleLinux/OL9/baseos/latest/$basearch/
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-oracle
gpgcheck=1
enabled=1
# dnf install oraclelinux-release-el9
# mv /etc/yum.repos.d/ol9-temp.repo /etc/yum.repos.d/ol9-temp.repo.disabled
The above is from https://yum.oracle.com/getting-started.html
The following is from https://gavinsoorma.com.au/knowledge-base/install-oracle-23c-free-developer-release-via-linux-rpm/ with minimal change ("23c" changed to “23ai”, DB_DOMAIN set to LOCAL)
# yum -y localinstall oracle-database-free-23ai-1.0-1.el9.x86_64.rpm
# mkdir u01
# cd /u01
# mkdir oracle
# cd oracle
# mkdir oradata
# cd /u01
# chown -fR oracle:oinstall oracle
# vi /etc/sysconfig/oracle-free-23ai.conf
and set LISTENER_PORT=1522, CHARSET=AL32UTF8, DBFILE_DEST=/u01/oracle/oradata, DB_DOMAIN=LOCAL, and SKIP_VALIDATIONS=false
# /etc/init.d/oracle-free-23ai configure
...
Copying database files
8% complete
[FATAL] ORA-03113: end-of-file on communication channel
29% complete
100% complete
[FATAL] ORA-03113: end-of-file on communication channel
7% complete
0% complete
Look at the log file "/opt/oracle/cfgtoollogs/dbca/FREE/FREE2.log" for further details.
Database configuration failed. Check logs under '/opt/oracle/cfgtoollogs/dbca'.
There are a few log files in /opt/oracle/cfgtoollogs/dbca/FREE. The biggest is trace.log_<timestamp>. But none seems to tell me what caused ORA-03113. Any help is appreciated.
[Additional finding]
In /opt/oracle/diag/rdbms/free/FREE/trace, I find alert_FREE.log. The only complaint in there is no HugePages. So I allocated 800 HugePages, which is bigger than the required 766. The other trace file, FREE_ora_998505.trc, may be relevant:
*** 2024-12-13T13:16:05.792709-06:00
WARNING: ORA-27125 during segment allocation: (size 6144K, pagesize 2048K)
WARNING: ORA-27125 during segment allocation: (size 16M, pagesize 2048K)
WARNING: ORA-27125 during segment allocation: (size 6144K, pagesize 2048K)
SGA (id: 1) creation time at startup: 4 ms
I:2024-12-13 13:16:05.830 : [ GPNP][ GPNP] clsgpnpkwf_checkDir: failed SldirOpen(/opt/oracle/product/23ai/dbhomeFree/gpnp/a1trldbms7/), ret=-5
I:2024-12-13 13:16:05.832 : [ GPNP][ GPNP] clsgpnpkwf_checkDir: failed SldirOpen(/opt/oracle/product/23ai/dbhomeFree/gpnp/a1trldbms7/), ret=-5
I:2024-12-13 13:16:05.832 : [ GPNP][ GPNP] clsgpnpkwf_initwfloc: [at clsgpnpkwf.c:665] Result: (7) CLSGPNP_IO. (:GPNP01004:)Fatal: Cannot open neither PEER nor PRDR GPnP wallet. No more wallets to verify GPnP configuration (profile). Check GPnP configuration: wallet(s) either missing or do not have access privileges. statres: -5, Wallet home : /u01/app/grid/gpnp/a1trldbms7/.
…
I:2024-12-13 13:16:05.846 : [ GPNP][ GPNP] clsgpnp_Init init failed. Error: CLSGPNP_ERR (1) .
I:2024-12-13 13:16:05.854 : [ default][ default] clsu_get_oda_type(): scls_scr_create() failed, ret2=1 fileName=odatype userName=oracle error=SLOS: cat=-1, opn=(none), dep=(none), loc=scrcreate5
need ha priv
In /opt/oracle/product/23ai/dbhomeFree, there's no gpnp. Also, “need ha priv” seen at the end, according to Doc ID 1634671.1, is due to node name change. But we never changed node name.