RAC on OCFS2 acceptance testing

From: Boris Dali <boris_dali_at_yahoo.ca>
Date: Wed, 27 Dec 2006 11:27:38 -0500 (EST)
Message-ID: <658359.98818.qm@web32412.mail.mud.yahoo.com>

I have an opportunity to conduct a basic technical acceptance testing, lasting no more than 5 days, on a pre-prod environment. It consists of a 2-node 10gR2 patchset 1 RAC database hosted on SLES 9 update 3 boxes with everything residing on OCFS2. And when I say "everything" I mean everything: online redo, archives, control files, datafiles, OCR, voting disks. The only exceptions are the Oracle binaries and the admin stuff (bdump, udump...) – local to each node (I know, I know... decision was made before my time – it’s not Ferrari, but performance is not my concern at the moment). There are also 2 standbys, whereas one is configured as a target for the 30second FSFO
(fast-start failover) and the 2-nd is just a
"bystander" standby (to be activated, obviously disabling FSFO, only as a last resort).

Tentative schedule for the testing week comprises the following test categories (the last two categories are not included in this tests as the backup infrastructure is not fully setup yet and God knows what app people are thinking...):

1. FO: Single primary failure [instance abort, SIGKILL listener, crs, cssd, power off a node, interconnect NIC failure (not bonded), HBA failure – how? ]
2. FO: Interconnect [FC switch is unfortunately shared with SAP, so powering it off is not an option L (and yes, there’s only one switch) ]
3. FO: Both primary instances/listeners/nodes failures [same as 1, but near simultaneous on both nodes]
4. FO: Cascading failures [?]
5. DR [srvctl stop database –d <…> -o abort and FSFO to a target standby under 2min, change the threshold and manual FO and switchback, FO to a non-target standby]
6. Load test [basic built-in synthetic DG 5-hour load of aprox. 3 MB/s, 10 tps – verify LB (server-side TAF, no FAN)]
7. Recovery [Loss of database, ocr (single copy and all), voting disks (single copy and all), spfile, cf
   (single copy and all), online redo
   (non-current/active, active, current – all multiplexed
   copies), binary tree, physical and logical corruptions
   (e.g. single table block corruptions with bbed), etc.
   
8. Application load and stress tests (qaload?)

Any evil thoughts (but not evil enough to kill the system completely, as I have to hand it over at the end of a week)?

Thanks,
Boris Dali.

---

Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

--
http://www.freelists.org/webpage/oracle-l

Received on Wed Dec 27 2006 - 10:27:38 CST