Fact: Oracle (all versions) and FC5 are completely incompatible because
of the new
glibc. When ones enter "FC5 Oracle10.2" into Google, the first thing
that pops out
is the following article:
http://www.oracle-base.com/articles/...nOnFedora5.php

This article leads to the conclusion that there is no problem with
installing Oracle 10.2
on FC5. Nothing could be further from the truth. This was just a "draft
article" written
before the final version of FC5. This final version has new glibc which
is incompatible
with Oracle 10.2 at the moment. This article is severely misleading.
Please, do not
follow the advice and do not attemt to use FC5 as an Oracle RDBMS
server. It cannot
be done at the moment.
I looked through the article for an email address of the proud author
(someone known
as "Tim") but I couldn't find it. Therefore, I decided to publish my
warning here.
Look instead on OTN and on this group. OTN Forums for Linux and
Installation have
some good descriptions of the problem.

"Bart the bear" <bartthebear@gmail.com> wrote in message
news:1143569190.948092.52150@z34g2000cwc.googlegro ups.com...
> Fact: Oracle (all versions) and FC5 are completely incompatible because
> of the new
> glibc. When ones enter "FC5 Oracle10.2" into Google, the first thing
> that pops out
> is the following article:
> http://www.oracle-base.com/articles/...nOnFedora5.php
>
> This article leads to the conclusion that there is no problem with
> installing Oracle 10.2
> on FC5. Nothing could be further from the truth. This was just a "draft
> article" written
> before the final version of FC5. This final version has new glibc which
> is incompatible
> with Oracle 10.2 at the moment. This article is severely misleading.
> Please, do not
> follow the advice and do not attemt to use FC5 as an Oracle RDBMS
> server. It cannot
> be done at the moment.
> I looked through the article for an email address of the proud author
> (someone known
> as "Tim") but I couldn't find it. Therefore, I decided to publish my
> warning here.
> Look instead on OTN and on this group. OTN Forums for Linux and
> Installation have
> some good descriptions of the problem.
>


I see there is a facility for adding comments at
the bottom, under the link:

2 comments, read/add them...

so you could append your comment
there. It looks like the author does respond
to comments.


--
Regards

Jonathan Lewis
http://www.oracle.com/technology/com...ce1.html#lewis

The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html

Cost Based Oracle: Fundamentals
http://www.jlcomp.demon.co.uk/cbo_book/ind_book.html

Point taken. Obviously, I didn't look hard enough.

On Tue, 28 Mar 2006 10:06:31 -0800, Bart the bear wrote:

> Look instead on OTN and on this group. OTN Forums for Linux and
> Installation have
> some good descriptions of the problem.

Well, as an author of one of those technical descriptions, here it is:

************************************************** ************************
Registered: 3/28/99

Re: Oracle 10g R2 on FC5; ORA-12157 error on installation
Posted: Mar 28, 2006 8:42 AM in response to: mgogala in response to:
mgogala
Click to reply to this thread Reply

The problem is in Linux threads: the new GLIBC 2.4.4-4 doesn't have Linux Threads:
https://www.redhat.com/archives/fedo.../msg00217.html
FC5 replaced it with the new thread library implementation called "NPTL".
As a result, applications which try to use the old threading methods will experience
"failure to communicate" expressed, of course, as 12157.

One possible solution is here:
http://lists.matureasskickers.net/pi...ch/000181.html

Oracle Corp. is probably recompiling its software with new NPTL library and gcc-4.1
as we speak but it will probably not appear until RH EL5 (FC5 is a "testing ground"
for RH EL5). So, you should either downgrade to FC4 or something like CentOS or
wait until Oracle for RH EL5 hits the road. Personally, I'm downloading CentOS and
swearing at Red Hat. You can nail your NIC to a wall, it wan't change a thing. The
problem is in the new GLIBC.
I will, however, keep trying to hack the present installation until the weekend.
Any news will be promptly reported.
************************************************** ************************

Since then, I downloaded an earlier glibc and unpacked it first in a
separate directory and, after that, in $ORACLE_HOME/lib/i686. It didn't
work.
Basically, FC5 intentionally broke the compatibility with the old
code, without telling anybody, just to get users to go to FC5. I will
switch from FC5 to CentOS. This is the way to have a supported version
of Oracle and inflict the maximal damage on Red Hat. Not only will I not
be a free tester for their product, I will use the enterprise version
for free. That is the minimum that I can do after what Red Hat has put
me through.

$ md5sum -c md5sum.dvd.asc
CentOS-4.3-i386-binDVD.iso: OK

It accepts any rpm for RH EL4 and will do just fine. What annoys me
the most is the way Red Hat has done it: no application testing,
no announcements, nothing. FC5 was designed as a trap waiting for
the innocent "testers" to fall into it. All the features that are in
FC5 now will be in RH EL5 in 6 months and, consequently, in the next
CentOS clone. After this experience with FC5, my advice for anybody
who would use Fedora for Oracle server is to reconsider. You will be an
unpaid tester for the vendor who doesn't even value enough to warn
you about the problem. As for the article mentioned in the beginning of
the thread, I read it before downloading FC5 and was convinced that
everything will be OK. The article is seriously misleading and should be
withdrawn with an apology to all those who lost some serious time
believing into claims made by the author. A week of FC5 actually being
in the wild was more then enough for the author to re-test and correct
his claims. When this wasn't done, warning on this group is completely
in order. Nobody else needs to lose sleep over @#$%! Fedora Core 5.


--
http://www.mgogala.com

On Tue, 28 Mar 2006 19:14:30 +0100, Jonathan Lewis wrote:

> I see there is a facility for adding comments at
> the bottom, under the link:
>
> 2 comments, read/add them...
>
> so you could append your comment
> there. It looks like the author does respond
> to comments.

The article is seriously misleading. I read it before downloading FC5 and
losing some serious time on configuration and hacking it. The author
should have retracted it by now. He still hasn't done so, therefore a
warning here is in order and will, probably, lower the number of possible
victims.

--
http://www.mgogala.com

"Mladen Gogala" <gogala@sbcglobal.net> wrote in message
newsan.2006.03.29.02.09.42.764132@sbcglobal.net. ..
> On Tue, 28 Mar 2006 10:06:31 -0800, Bart the bear wrote:
>
>> Look instead on OTN and on this group. OTN Forums for Linux and
>> Installation have
>> some good descriptions of the problem.
>
> Well, as an author of one of those technical descriptions, here it is:
>
> ************************************************** ************************
> Registered: 3/28/99
>
> Re: Oracle 10g R2 on FC5; ORA-12157 error on installation
> Posted: Mar 28, 2006 8:42 AM in response to: mgogala in response to:
> mgogala
> Click to reply to this thread Reply
>
> The problem is in Linux threads: the new GLIBC 2.4.4-4 doesn't have Linux
> Threads:
> https://www.redhat.com/archives/fedo.../msg00217.html
> FC5 replaced it with the new thread library implementation called "NPTL".
> As a result, applications which try to use the old threading methods will
> experience
> "failure to communicate" expressed, of course, as 12157.
>
> One possible solution is here:
> http://lists.matureasskickers.net/pi...ch/000181.html
>
> Oracle Corp. is probably recompiling its software with new NPTL library
> and gcc-4.1
> as we speak but it will probably not appear until RH EL5 (FC5 is a
> "testing ground"
> for RH EL5). So, you should either downgrade to FC4 or something like
> CentOS or
> wait until Oracle for RH EL5 hits the road. Personally, I'm downloading
> CentOS and
> swearing at Red Hat. You can nail your NIC to a wall, it wan't change a
> thing. The
> problem is in the new GLIBC.
> I will, however, keep trying to hack the present installation until the
> weekend.
> Any news will be promptly reported.
> ************************************************** ************************
>
> Since then, I downloaded an earlier glibc and unpacked it first in a
> separate directory and, after that, in $ORACLE_HOME/lib/i686. It didn't
> work.
> Basically, FC5 intentionally broke the compatibility with the old
> code, without telling anybody, just to get users to go to FC5. I will
> switch from FC5 to CentOS. This is the way to have a supported version
> of Oracle and inflict the maximal damage on Red Hat. Not only will I not
> be a free tester for their product, I will use the enterprise version
> for free. That is the minimum that I can do after what Red Hat has put
> me through.
>
> $ md5sum -c md5sum.dvd.asc
> CentOS-4.3-i386-binDVD.iso: OK
>
> It accepts any rpm for RH EL4 and will do just fine. What annoys me
> the most is the way Red Hat has done it: no application testing,
> no announcements, nothing. FC5 was designed as a trap waiting for
> the innocent "testers" to fall into it. All the features that are in
> FC5 now will be in RH EL5 in 6 months and, consequently, in the next
> CentOS clone. After this experience with FC5, my advice for anybody
> who would use Fedora for Oracle server is to reconsider. You will be an
> unpaid tester for the vendor who doesn't even value enough to warn
> you about the problem. As for the article mentioned in the beginning of
> the thread, I read it before downloading FC5 and was convinced that
> everything will be OK. The article is seriously misleading and should be
> withdrawn with an apology to all those who lost some serious time
> believing into claims made by the author. A week of FC5 actually being
> in the wild was more then enough for the author to re-test and correct
> his claims. When this wasn't done, warning on this group is completely
> in order. Nobody else needs to lose sleep over @#$%! Fedora Core 5.
>
>
> --
> http://www.mgogala.com
>

While your compaints are legit, Red Hat should not be the one to blame.
After all it is not their web site. Any time you want to experiment with
unsupported combinations, you are at your own risk, unfortunately.

On Wed, 29 Mar 2006 03:55:26 +0000, Bob Jones wrote:

> While your compaints are legit, Red Hat should not be the one to blame.
> After all it is not their web site. Any time you want to experiment with
> unsupported combinations, you are at your own risk, unfortunately.

Well, it was a tacit understanding that things will work with some
hacking, if needed. I successfully installed various versions of Oracle on
FC2,FC3 & FC4. I expected it to continue with FC5, especially after that
misleading article. This is the first version completely and utterly
incompatible with Oracle (I never touched FC1, so I don't know about that).
Red Hat also deserves a full share of blame for idiotic things like this:

[root@medo ~]# newaliases
hash map "Alias0": unsafe map file /etc/aliases.db: Permission denied
WARNING: cannot open alias database /etc/aliases
Cannot create database for alias file /etc/aliases
[root@medo ~]# whoami
root
[root@medo ~]#

Of course, makemap no longer works, as it says in the manual:

Notice: do not use makemap to create the aliases data base, because
newaliases puts a special token into the data base that is required by
sendmail.

FILES
/etc/aliases The mail aliases file

So, may annoyance with FC5 is multi-pronged. Also, WEP module doesn't work
with orinoco_plx. The card works, it gets the IP address from the DHCP
server and stops working. When I remove WEP, the card magically starts
working. (Of course I have MAC filtering, too).

All things considered, FC5 is a very, very disappointing version. The new
compiler breaks linux-wlan, NVIDIA driver needs heavy patching and
pre-stable kernel, while OSS sound software cannot be installed. Do they
test the software or they just let us test it? Well, they just lost one
"tester".

--
http://www.mgogala.com

To "Bart the bear":

The article you are referring to was clearly labelled as a draft
article. You only know it exists because Google has indexed it and made
it available on a Google search.

I never claimed it worked against the final version of FC5, afterall,
it was written prior to the final release.

May I suggest in future:

1) You read my articles fully, then you will not be lead down the
garden path by "draft" documents.

2) You take time to examine my site, where you will find a "comments"
section at the end of each article, a "forum" and a "contact
ORACLE-BASE.com" section. Most people seem capable of finding some way
of contacting me directly.

3) If you had taken more than a cursory glance at the site, you would
have seen a "Site Info" page that explains who I am (someone known as
"Tim") and why I do the site. To summarize, I do it for me! If you like
it fine! If you don't, it's no skin off my nose!

4) After someone has hand-held you through the process of contacting
me, you could attempt to leave a more pleasant message. You know
nothing about me, but I kow a lot about you now

Cheers

Tim...

PS. I've amended the comments at the start of the article. I assume you
will no longer fall into the the trap of thinking that "DRAFT" means
"Finished and working" now!

Mladen:

"Serious Misleading" - No! It was a draft article and it clearly stated
so! Misleading would be if I had claimed it was a finished article that
worked against the production release of FC5. I did not!

"The author should have retracted it by now." - It's a draft. Why
should I retract a draft article. It does not, and never did, claim to
be anything it is not.

"still hasn't done so" - Are you yanking my chain? I'll give you a time
line:

1) I write the "draft" article ages ago!
2) I get a comment from Bart at "2006-03-28 20:06:52".
3) I get a comment from you at "2006-03-29 04:19:46".
4) I check my mail this morning and see both comments for th first
time.
5) I waste time replying to your pointless belly aching because you
guys don't have the presence of mind to understand the phrase "THIS IS
A DRAFT DOCUMENT" means something different to "This is a complete and
fully functional example".
6) I add some stronger comments to the start of the document!

Perhaps I should add a draft point (7):

7) "Draft": I actually download the final release of FC5 and test my
"draft" articles. By the way, there are two of them. Perhaps you need
to start a thread complaining about the second one also!

It is still less that 24 hours since the original comment was made and
I've made my response. I think this is pretty darn quick.

Perhaps you guys should think twice before yo start criticising other
people. In my opinion, the way you've handled this issue is completely
unreasonable.

Get a life!

Tim...

On Wed, 29 Mar 2006 02:54:58 -0800, Tim... wrote:

> "Serious Misleading" - No! It was a draft article and it clearly stated
> so! Misleading would be if I had claimed it was a finished article that
> worked against the production release of FC5. I did not!

I wouldn't have attempted the installation, had I read anything about the
problems. There is much more serious problem then few changes to white
space or new line in gennttab.

--
http://www.mgogala.com