Re: oracle rac 10g and OS clusterware.

  • From: Dan Norris <dannorris@xxxxxxxxxxxxx>
  • To: john.hallas@xxxxxxxxxx, ujang.jaenudin@xxxxxxxxx, Peter McLarty <peter.mclarty@xxxxxxxxxxxxxxxxxx>
  • Date: Wed, 5 Dec 2007 18:36:17 -0800 (PST)

Thanks for the love, John :).

In my experience, node eviction only takes place when there's a component 
failure in the cluster. From what I've seen, that's almost always a hardware 
failure and I'd characterize it as "rare". However, as I'm sure someone will 
point out if I don't say it, if it happens just one time ever, that's all it 
takes to cause a very big problem if the cluster doesn't detect the failure 
properly and handle it properly (and quickly). While I'm sure there are plenty 
of stories about unpleasant experiences with Oracle Clusterware's method of 
node eviction, I'd submit that if you look hard enough, you're always bound to 
find someone with a bad experience about something that you think is very 
solid. 

The short answer to your question is that I think node eviction is a rare 
occurrence on most "normal" configurations. 

Dan

----- Original Message ----
From: John Hallas <john.hallas@xxxxxxxxxx>
To: dannorris@xxxxxxxxxxxxx; ujang.jaenudin@xxxxxxxxx; Peter McLarty 
<peter.mclarty@xxxxxxxxxxxxxxxxxx>
Cc: oracle-l@xxxxxxxxxxxxx
Sent: Wednesday, December 5, 2007 8:21:09 AM
Subject: RE: oracle rac 10g and OS clusterware.





 
 


<!--
 _filtered {font-family:Tahoma;panose-1:2 11 6 4 3 5 4 4 2 4;}
 _filtered {font-family:"Comic Sans MS";panose-1:3 15 7 2 3 3 2 2 2 4;}
/* Style Definitions */
 p.MsoNormal, li.MsoNormal, div.MsoNormal
        {margin:0cm;margin-bottom:.0001pt;font-size:12.0pt;font-family:"Times 
New Roman";}
a:link, span.MsoHyperlink
        {color:blue;text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
        {color:blue;text-decoration:underline;}
span.EmailStyle17
        {font-family:"Comic Sans 
MS";color:blue;font-weight:normal;font-style:normal;text-decoration:none none;}
 _filtered {margin:72.0pt 90.0pt 72.0pt 90.0pt;}
div.Section1
        {}
-->






Good response
Dan. How often does node eviction take place on a ʽnormalʼ
configuration (if there is such a thing). 
 

  
 

PS I came
across an excellent presentation of yours this morning on HA options for Oracle

 

http://www.dannorris.com/wp-content/uploads/2007/11/ha-options-for-oracle-db-oow2007-slides.pdf
 

  
 

John
 

  
 

  
 










From:
oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf 
Of Dan Norris

Sent: 05 December 2007 03:32

To: ujang.jaenudin@xxxxxxxxx; Peter
McLarty

Cc: oracle-l@xxxxxxxxxxxxx

Subject: Re: oracle rac 10g and OS
clusterware.
 




  
 





Ujang,



Oracle Clusterware 10g will always handle node fencing (aka eviction) by
rebooting one or more nodes. Oracle didn't have it's own cluster manager
software for most platforms on 9i. So, if your customer was on a platform where
Oracle didn't have a cluster manager of its own (it only had Linux and
Windows), then building the RAC cluster would have required some 3rd party
software. Some of the 3rd party cluster managers used different schemes to
fence nodes from the cluster. Many of them used I/O fencing which was somewhat
less disruptive in that it didn't require a node reboot, but required more
proprietary interfaces to be used to access and manage storage. 



The answer to your question is that it depends. You will always have to have
Oracle Clusterware if you're building a 10g cluster. However, if you
additionally use a 3rd party clusterware (one that uses I/O fencing instead of
reboots to handle node eviction), Oracle Clusterware will not handle cluster
membership and therefore won't impose its "reboot the other node"
scheme of node eviction. I should mention that I believe this is true in most
cases, but not all cases. 



If you use just Oracle Clusterware (which is the only requirement and handles
all cluster management needs albeit with using node reboots to evict them from
the cluster), you'll be subject to node reboots if node eviction is required. 



See Kirk McGowan's post on Oracle Clusterware's node eviction methods at 
http://blogs.oracle.com/kmcgowan/2007/08/09#a13
for some additional insight. 



Dan
 








 




  
 















The information included in this email and any files transmitted with it may 
contain information that is confidential and it must not be used by, or its 
contents or attachments copied or disclosed, to persons other than the intended 
addressee. It must not be used by, or its contents or attachments copied or 
disclosed to, any persons other than the intended addressee.  If you have 
received this email in error, please notify BJSS.

In the absence of written agreement to the contrary BJSSʼ relevant standard 
terms of contract for any work to be undertaken will apply.

Please carry out virus or such other checks as you consider appropriate in 
respect of this email.  BJSS do not accept responsibility for any adverse 
effect upon your system or data in relation to this email or any files 
transmitted with it.

BJSS Limited, a company registered in England and Wales (Company Number 
2777575), VAT Registration Number 613295452, Registered Office Address, First 
Floor, Coronet House, Queen Street, Leeds, LS1 2TW










Other related posts: