Skip to Main Content
Forums

Oracle Database Discussions

Announcement

For appeals, questions and feedback about Oracle Forums, please email oracle-forums-moderators_us@oracle.com. Technical questions should be asked in the appropriate category. Thank you!

Install Oracle DB on unsupported Linux - Ora-00600 [16305] during startup.

dlackaDec 27 2020

Hello,
I'm trying to install Oracle 12.1.0.2 (EE) on Debian 10 (Buster) for some learning/testing purpose. runInstaller seems finished without error (incl. "Linking binaries"), but dbca (or startup nomount... "by hand") fails with ORA-00600 [16305] error. googling for a while I've found the folllowing (Doc ID 1928580.1)
ORA-600 [16305] During Startup (0 Bytes)Unfortunately I've no MOS access. Would sb be so kind to give me a hint or reveal the secret under "Cause"/"Solution" (is it a bug or just mis-config I've made)?
Many thanks in advance,
Leslie
(Pls. No "use VM, OL, RHEL... etc." hints. Thanks)
This post has been answered by Billy Verreynne on Dec 29 2020
Jump to Answer

Comments

Billy Verreynne

Sticky bit for /tmp is not set. Use chmod 1777 /tmp and relink.
By the way, the shared object library libgcc_s.so on newer Debian kernels fail to load correctly for Oracle 18c. Have not checked 19c, or tried 12c.
If you do get a fatal error starting Oracle, run it via a strace to have a look see.

dlacka

Hello Billy,
Thank you for your answer. The sticky bit is factory default on /tmp.
drwxrwxrwt 1 root root 874 Dec 28 20:15 tmp
Excerpt from trace:
System name: Linux
Release: 5.9.0-0.bpo.2-amd64
Version: #1 SMP Debian 5.9.6-1~bpo10+1 (2020-11-19)
Machine: x86_64
kmmingb()+172 call ksesic0() 7FD8D5569CA0 ? 000040000
0FFFFFFFF 0FFFFFFFF
7FD8D4DDA000 000000000
kmmsgini()+15 call kmmingb() 7FD8D5569CA0 ? 000040000 ?
0FFFFFFFF ? 0FFFFFFFF ?
7FD8D4DDA000 ? 000000000 ?
opistr_real()+1229 call kmmsgini() 7FD8D5569CA0 ? 000040000 ?
0FFFFFFFF ? 0FFFFFFFF ?
7FD8D4DDA000 ? 000000000 ?
opistr()+491 call opistr_real() 000000030 000000002
7FFDB290B1F8 0FFFFFFFF ?
7FD8D4DDA000 ? 000000000 ?
opiodr()+1165 call opistr() 000000030 000000002
7FFDB290B1F8 0FFFFFFFF ?
7FD8D4DDA000 ? 000000000 ?
ttcpip()+2699 call opiodr() 000000030 000000002
7FFDB290B1F8 ? 0FFFFFFFF ?
7FD8D4DDA000 ? 000000000
opitsk()+1734 call ttcpip() 7FD8D557F090 ? 000000030 ?
7FFDB290B1F8 000000000 ?
7FFDB290AC58 7FFDB290B404
opiino()+945 call opitsk() 000000900 000000000
7FFDB290B1F8 ? 000000000 ?
7FFDB290AC58 ? 7FFDB290B404 ?
opiodr()+1165 call opiino() 00000003C 000000004
7FFDB290C898 000000000 ?
7FFDB290AC58 ? 7FFDB290B404 ?
opidrv()+587 call opiodr() 00000003C 000000004
7FFDB290C898 ? 000000000 ?
7FFDB290AC58 ? 000000000
sou2o()+145 call opidrv() 00000003C 000000004
7FFDB290C898 000000000 ?
7FFDB290AC58 ? 000000000 ?
opimai_real()+154 call sou2o() 7FFDB290C870 00000003C
000000004 7FFDB290C898
7FFDB290AC58 ? 000000000 ?
ssthrdmain()+412 call opimai_real() 000000000 7FFDB290CB80
000000004 ? 7FFDB290C898 ?
7FFDB290AC58 ? 000000000 ?
[...]
========== FRAME [16] (kmmingb()+172 -> ksesic0()) ==========
defined by frame pointers 0x7ffdb29073a0 and 0x7ffdb2907300
CALL TYPE: call ERROR SIGNALED: yes COMPONENT: (null)
RDI 00007FD8D5569CA0 RSI 0000000000040000 RDX 00000000FFFFFFFF
RCX 00000000FFFFFFFF R8 00007FD8D4DDA000 R9 0000000000000000
RAX 0000000000000001 RBX 00007FFDB2907510 RBP 00007FFDB29073A0
R10 0000000000000000 R11 0000000000000000 R12 00007FD8D5569CA0
R13 00007FD8D559B280 R14 00007FD8D559B280 R15 0000000000003FB1
RSP 00007FFDB2907310 RIP 0000000002430E7C
Dump of memory from 0x7ffdb2907300 to 0x7ffdb29073a0
I'm going to investigate on your 2nd hint (libgcc_s.so).
Thanks,
Leslie

Billy Verreynne
Answer

I have Oracle XE 11g running on Debian Ubuntu without issues. Could not get XE 18c to run. Segment violations.
That support note said the sticky bit is not set - which likely means whatever triggered that error on your side is an unknown from a support note perspective.

Marked as Answer by dlacka · Dec 29 2020
dlacka

Hello Billy,
Your 2nd hint kicked me to the right direction. I don't know the exact source of problem yet (honestly: don't care), but the problem solved using a little "downgrade" .
Reinstall 10.5 Buster (instead of 10.7), with Kernel 4.19.0-10 (instead of 5.9).
Install Oracle 12c (Limits, packages, softlinks as usual) and dbca (startup nomount...) started with no errors.
Thank you for the "spark" again (and Happy New Year, without bio and tech viruses).
Leslie

Billy Verreynne

Awesome. I suspected something like this, but did not want to do the downgrade path to confirm.

1 - 5

Post Details

Added on Dec 27 2020
#database-installation
5 comments
140 views