Back to normal life, for a few days…

I’m back in the office today, recovering from the disaster recovery week 🙂

One of my friends phoned today and asked me to cover his yoga classes tonight, so I’ll have to give Karate a miss. Shame, but a friend in need etc.

I hope he feels better by the weekend. We’re both going to a Yoga course in Edinburgh on Friday and I don’t want him sneezing on me during the flight 🙂

Cheers

Tim…

Disaster Recovery Testing Update…

So far it’s been a very long and very tiring week. Lots of standing up in very cold server rooms.

The disaster recovery testing has highlighted quite a few interesting points about our kit and our backup and recovery approach. I think this session will stand us in good stead for the second round of testing in a few weeks. By then I’m sure we will have ironed out some of the creases 🙂

One more day to go, then home…

Cheers

Tim…

Disaster Recovery Testing and stuff…

I’m off to HP for the rest of the week to help test our disaster recovery strategy, so I guess things will be a little quiet here for a few days.

OEM Grid Control Update:
So far things look pretty good. The agents still use a stack of memory, just under 1 Gig resident for the Grid Agent and IAS Console together on a Release 1 AS 10g server. Personally I think this is more than a little excessive, but the CPU load isn’t too bad at all, not like the previous version…

I knocked up a quick example of DBMS_PIPE in answer to a question on my forum. Personally, I use AQ for all my messaging these days, but each to their own 🙂

Cheers

Tim…

Hyperthreading Hurts Server Performance?

I saw this article today: Hyperthreading Hurts Server Performance?

We use lots of commodity hardware running Linux for our application servers. It would be interesting to do the Pepsi challenge and see if HT is affecting our performance…

Cheers

Tim…

DBMS_ASSERT and stuff…

I saw a post about the DBMS_ASSERT package on Pete Finnigan’s Oracle security weblog and I couldn’t find any documentation about it, so I thought I’d have a little play and write something about it:

DBMS_ASSERT – Sanitize User Input to Help Prevent SQL Injection

I’m not sure I’ll find a use for it, but it’s always good to know what new bits have been added 🙂

This made me laugh: Counterfeiters send jammed printer for repair

Cheers

Tim…

Colour Schemes…

Howard Rogers made a point about colour scheme selection in reply to my Website Design post. My site is already CSS compliant, so I thought I might as well add a colour selector on the bottom-right of the homepage.

I’m not the most artistic of people, so if anyone has any good colour combinations feel free to post them 🙂

Cheers

Tim…

Oracle 10g Release 2 Grid Control Installation On Red Hat Enterprise Linux and CentOS…

I’ve been having a play with the latest version of the Grid Control:

Oracle 10g Release 2 Grid Control Installation On Red Hat Enterprise Linux and CentOS

It installed no problem and appears to be working right out of the box.

The big question is, does it soak up all the CPU and memory on my servers?

The first version of the Grid Control was a complete nightmare for me. The agents worked fine on the database nodes, but they gobbled up loads of memory and CPU on the application servers, making them unusable. I raised a TAR and was told it was a bug that was fixed in the next release.

I’ve only installed an agent on one development application server at the moment. I’ll add a few more to non-critical servers tomorrow and see how it gets on. Fingers crossed! I’ll post the outcome of this test in a few days.

Cheers

Tim…

Setting the record straight…

In my previous post (Support goes on, and on, and on…) I mentioned that the latest version of the patch in question was issued with some missing directories. Infact, this was incorrect.

I don’t know how, but I was able to see the new directory structure with the old readme file. I can only assume they my text editor, which has remote editing via FTP, was having a bad day. I hear a chorus of, “You should have used vi!”, coming.

Anyway, the latest version of the patch applied successfully to both the development and test environments, so I’m going to schedule it for production some time soon.

My sentiments over the last post remain unchanged, but certainly the last incident was in no way the fault of Oracle Support. I thought it was only fair to set the record straight!

Cheers

Tim…

Support goes on, and on, and on…

Just to set the scene:

04-APR-05 – I opened a TAR because the RACGIMON process was taking 20-40% of the CPU on the cluster root node of our Tru64 production RAC (10.1).

05-SEP-05 – A backport of a fix in 10.1.0.5.0 was produced. Note. 10.1.0.4.0 is still not released for Tru64. The patch failed to install on either the development or test systems, so I didn’t go near the production one.

24-OCT-05 – It was decided that the installation documentation was incorrect and the patch was reissued.

12-NOV-05 – The reissued patch is missing some directories. As a result the installation is impossible. (Update: Not Oracle’s fault. See Setting the record straight…)

Conclusion:

So far it’s been 7 months, during which time we’ve been loosing between one tenth an one fifth of our total processing power on our production system, and still no resolution. Great to know that Oracle are on top of their game… NOT!

Cheers

Tim…