Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Useful piece of software
43 views
Skip to first unread message
Mladen Gogala
Apr 29, 2012, 7:50:20 PM4/29/12
to
I am building my own RAC on my home desktop PC and, as expected, the
largest problem was with bind (name server). I was trying to fix it for
the whole afternoon, without success. I saw countless recipies, Tim Halls
among others, but still couldn't get it to work.
Then, I discovered dnsmasq:
http://www.thekelleys.org.uk/dnsmasq/doc.html
That is the name server for small home networks, which reads the hosts
file. Lo and behold, it took me only 10 minutes with it.
ns1@ns1-VirtualBox:~$ nslookup rac1.localdomain
Server: 127.0.0.1
Address: 127.0.0.1#53
Name: rac1.localdomain
Address: 192.168.1.150
ns1@ns1-VirtualBox:~$ nslookup scan.localdomain
Server: 127.0.0.1
Address: 127.0.0.1#53
Name: scan.localdomain
Address: 192.168.1.230
Name: scan.localdomain
Address: 192.168.1.220
Now, that's a useful thing, if anyone needs it.
--
http://mgogala.byethost5.com
largest problem was with bind (name server). I was trying to fix it for
the whole afternoon, without success. I saw countless recipies, Tim Halls
among others, but still couldn't get it to work.
Then, I discovered dnsmasq:
http://www.thekelleys.org.uk/dnsmasq/doc.html
That is the name server for small home networks, which reads the hosts
file. Lo and behold, it took me only 10 minutes with it.
ns1@ns1-VirtualBox:~$ nslookup rac1.localdomain
Server: 127.0.0.1
Address: 127.0.0.1#53
Name: rac1.localdomain
Address: 192.168.1.150
ns1@ns1-VirtualBox:~$ nslookup scan.localdomain
Server: 127.0.0.1
Address: 127.0.0.1#53
Name: scan.localdomain
Address: 192.168.1.230
Name: scan.localdomain
Address: 192.168.1.220
Now, that's a useful thing, if anyone needs it.
--
http://mgogala.byethost5.com
onedbguru
Apr 30, 2012, 3:33:43 PM4/30/12
to
Just an FYI, you will have problems is you leave the SCAN-address(es) in the /etc/hosts file... you can only have one "name" entry in that file - if you have more than one, it will always read the first one. Make sure that when you nslookup multiple times that the scan-ip changes... Also, even in a two-node RAC, it is still recommended to have 3 scan addresses.
Mladen Gogala
Apr 30, 2012, 4:50:54 PM4/30/12
to
On Mon, 30 Apr 2012 12:33:43 -0700, onedbguru wrote:
> You have that problem only if you don't know DNS/bind :) :)
Which is true: I don't know DNS/bind. That is precisely why I recommend
> You have that problem only if you don't know DNS/bind :) :)
dnsmasq: it's simple, intuitive and does the job.
>
> Just an FYI, you will have problems is you leave the SCAN-address(es) in
> the /etc/hosts file... you can only have one "name" entry in that file -
/etc/hosts file. I don't envision ever referencing them directly.
> if you have more than one, it will always read the first one. Make sure
> that when you nslookup multiple times that the scan-ip changes... Also,
> even in a two-node RAC, it is still recommended to have 3 scan
> addresses.
RAC and the listener will then dispatch the connection to the appropriate
service. What would be the purpose of having 3 addresses for 2 nodes? Who
has made that recommendation and why? I don't really want to do much with
the SCAN listener, but Oracle mandates it, so I had to use shortcuts. I
am not going to become a network administrator in order to install RAC.
--
http://mgogala.byethost5.com
Noons
May 4, 2012, 3:08:03 AM5/4/12
to
On May 1, 6:50 am, Mladen Gogala <gogala.mla...@gmail.com> wrote:
> I am not going to become a network administrator in order to install RAC.
You, "dba 1.0" you...
> I am not going to become a network administrator in order to install RAC.
¦-)
Mladen Gogala
May 5, 2012, 1:43:06 AM5/5/12
to
doing remote operations, at least one node would crash. I ran the
installer on a single node, it was successful, but addNode.sh refused to
cooperate, saying that it doesn't know anything about the OS. I attempted
adding -IGNORESYSPREREQS to the command line, but to no avail, it was
attempting to run cluvfy. My conclusion is that 11.2.0.4 is in the cards
this fall or else it doesn't make sense to certify the complete line of
products against OL 6.x
I just downloaded OL 5.8 and hopefully, it will be easier to install.
--
http://mgogala.byethost5.com
Mladen Gogala
May 5, 2012, 8:21:15 PM5/5/12
to
On Sat, 05 May 2012 05:43:06 +0000, Mladen Gogala wrote:
> I had a problem with OL 6.2 and VirtualBox 4.1.4. When the installer was
> doing remote operations, at least one node would crash. I ran the
> installer on a single node, it was successful, but addNode.sh refused to
> cooperate, saying that it doesn't know anything about the OS. I
> attempted adding -IGNORESYSPREREQS to the command line, but to no avail,
> it was attempting to run cluvfy. My conclusion is that 11.2.0.4 is in
> the cards this fall or else it doesn't make sense to certify the
> complete line of products against OL 6.x I just downloaded OL 5.8 and
> hopefully, it will be easier to install.
Problem was in the type of the network interface used for the private
> I had a problem with OL 6.2 and VirtualBox 4.1.4. When the installer was
> doing remote operations, at least one node would crash. I ran the
> installer on a single node, it was successful, but addNode.sh refused to
> cooperate, saying that it doesn't know anything about the OS. I
> attempted adding -IGNORESYSPREREQS to the command line, but to no avail,
> it was attempting to run cluvfy. My conclusion is that 11.2.0.4 is in
> the cards this fall or else it doesn't make sense to certify the
> complete line of products against OL 6.x I just downloaded OL 5.8 and
> hopefully, it will be easier to install.
interconnect. I selected "internal" and it doesn't work. It must be
"bridged".
--
http://mgogala.byethost5.com
Noons
May 6, 2012, 4:11:54 AM5/6/12
to
Mladen Gogala wrote,on my timestamp of 6/05/2012 10:21 AM:
>> I had a problem with OL 6.2 and VirtualBox 4.1.4. When the installer was
>> doing remote operations, at least one node would crash. I ran the
>> installer on a single node, it was successful, but addNode.sh refused to
>> cooperate, saying that it doesn't know anything about the OS. I
>> attempted adding -IGNORESYSPREREQS to the command line, but to no avail,
>> it was attempting to run cluvfy. My conclusion is that 11.2.0.4 is in
>> the cards this fall or else it doesn't make sense to certify the
>> complete line of products against OL 6.x I just downloaded OL 5.8 and
>> hopefully, it will be easier to install.
>
> Problem was in the type of the network interface used for the private
> interconnect. I selected "internal" and it doesn't work. It must be
> "bridged".
>
>
>
Interesting. How are you finding Virtualbox 4.1.4? Solid? I'm thinking of
>> I had a problem with OL 6.2 and VirtualBox 4.1.4. When the installer was
>> doing remote operations, at least one node would crash. I ran the
>> installer on a single node, it was successful, but addNode.sh refused to
>> cooperate, saying that it doesn't know anything about the OS. I
>> attempted adding -IGNORESYSPREREQS to the command line, but to no avail,
>> it was attempting to run cluvfy. My conclusion is that 11.2.0.4 is in
>> the cards this fall or else it doesn't make sense to certify the
>> complete line of products against OL 6.x I just downloaded OL 5.8 and
>> hopefully, it will be easier to install.
>
> Problem was in the type of the network interface used for the private
> interconnect. I selected "internal" and it doesn't work. It must be
> "bridged".
>
>
>
upgrading to it but am too busy at the moment to go into "trial" mode.
Mladen Gogala
May 6, 2012, 11:42:11 AM5/6/12
to
On Sun, 06 May 2012 18:11:54 +1000, Noons wrote:
> Interesting. How are you finding Virtualbox 4.1.4? Solid? I'm
> thinking of upgrading to it but am too busy at the moment to go into
> "trial" mode.
As a matter of fact, it's extremely flaky. It's using host cache and I've
> Interesting. How are you finding Virtualbox 4.1.4? Solid? I'm
> thinking of upgrading to it but am too busy at the moment to go into
> "trial" mode.
had cases when it started swapping and woke up the OOM killer which then
aborted the VM. I am using Vbox on 64 bit Fedora 16,
--
http://mgogala.byethost5.com
Noons
May 7, 2012, 8:28:43 AM5/7/12
to
Mladen Gogala wrote,on my timestamp of 7/05/2012 1:42 AM:
> As a matter of fact, it's extremely flaky. It's using host cache and I've
> had cases when it started swapping and woke up the OOM killer which then
> aborted the VM. I am using Vbox on 64 bit Fedora 16,
Aye!.... Thanks for the heads up.
> As a matter of fact, it's extremely flaky. It's using host cache and I've
> had cases when it started swapping and woke up the OOM killer which then
> aborted the VM. I am using Vbox on 64 bit Fedora 16,
Mladen Gogala
May 7, 2012, 8:44:43 AM5/7/12
to
installation to the other virtual machines and the memory usage simply
explodes. I believe this to be a problem with the VM software.
--
http://mgogala.byethost5.com
Gunter Herrmann
May 7, 2012, 2:11:31 PM5/7/12
to
Hi!
Mladen Gogala wrote:
> Problem was in the type of the network interface used for the private
> interconnect. I selected "internal" and it doesn't work. It must be
> "bridged".
This bridged mode requires a processor and BIOS that support hardware virtualisation.
Some server manufacturers like Dell used cripled versions of some
processors only available to those manufacturers to criple their
low end systems. They wanted to sell Xeon processors.
A while ago we updated a Dell R200 server by replacing the original
Pentium (I think 5800) with a retail processor of the same type.
Suddenly hardware virtualization worked. (The retail processor was $80).
Gunter in Orlando, Fl.
Mladen Gogala wrote:
> Problem was in the type of the network interface used for the private
> interconnect. I selected "internal" and it doesn't work. It must be
> "bridged".
Some server manufacturers like Dell used cripled versions of some
processors only available to those manufacturers to criple their
low end systems. They wanted to sell Xeon processors.
A while ago we updated a Dell R200 server by replacing the original
Pentium (I think 5800) with a retail processor of the same type.
Suddenly hardware virtualization worked. (The retail processor was $80).
Gunter in Orlando, Fl.
0 new messages