Although Oracle Corp. hasn't certified yet the use of Oracle 19c software on OL9, I decided to try (at least) the installation of the Oracle Restart "Single Instance High Availability (SIHA)".
Expected message about (un)supported OS was ignored, in particular, by setting CV_ASSUME_DISTID in the shell before invoking ./gridSetup.sh :
$ export CV_ASSUME_DISTID=OL8.8
Then, after getting an error from the title, I simply scp'ed /usr/lib64/libc_nonshared.a over from another OL8 server, retried the linking, and the installation went on pretty well with one exception at the end - Oracle CVU thrown the error :
INFO: [Oct 6, 2023 3:08:33 PM] RPM Package Manager database ...FAILED (PRVG-13702)
INFO: [Oct 6, 2023 3:08:33 PM] Post-check for Oracle Restart configuration was unsuccessful.
INFO: [Oct 6, 2023 3:08:33 PM] Failures were encountered during execution of CVU verification request "stage -post hacfg".
INFO: [Oct 6, 2023 3:08:33 PM] RPM Package Manager database ...FAILED
INFO: [Oct 6, 2023 3:08:33 PM] PRVG-13702 : RPM Package Manager database files are corrupt on nodes "aaa".
The culprit were existing rpm packages in the system with obsolete SHA1 hash algorithm in their signatures. It worth to mention that this server was gradually upgraded from OL6 to OL9 through out of its life, so there were still some rpms signed by SHA1 signature which isn't supported anymore in OL9. The solution was to temporary implement the support of old (unsupported) signatures :
# update-crypto-policies --set LEGACY
and to check:
# update-crypto-policies --show
LEGACY
Next attempt of running CVU has finally succeeded and the installation process (including root scripts for the upgrade part) finished up without any errors.
I run :
# update-crypto-policies --set DEFAULT
to return the changed things back.
No comments:
Post a Comment