Well, I'm a "dinosaur" (aka, Grumpy Old Guy) and was trained that if the data was valuable enough to be put in a database in the first place, it should be treated that way; but, back when I was managing "real" databases for various companies; MINIUMUM Standard Operating Procedure after ANY patch was:
- Shut down the database (if it was running after the patch) and all other associated processes (like web servers or listeners)
- Relink Oracle Home
- Restart Oracle and all other processes
Of course, this leaves out a lot of steps that you might need to do BEFORE patching (like read, mark, learn and inwardly digest the release notes), or in the course of patching, like all the "catproc" stuff and any troubleshooting (for example, when the relink fails). We also did "the usual" things to mitigate the downtime required, including OS Clusters (before RAC) and RAC, or DataGuard, or...., depending on the environment and business requirements.
MOST of the time, the "powers that be" listened when I asked, "What kind of outage would you prefer? A scheduled, coordinated one of a 'time certain' duration (and we usually beat that); or a 'fire drill' of uncertain impact and duration when something breaks". Occasionally someone thought "they knew better" and had to "learn by experience".
At one particular employer, who shall remain nameless, when I arrived, they had not had a patch "succeed on the first try" in 4 years. That changed, not on the first or even second try, but, over time.
(And by the way, this "Standard Operating Procedure" was born of "knowing better" and occasionally, but less frequently over time, being "reminded" why "Good judgement comes from experience. Experience comes from bad judgement").
Good luck!
Clay Jackson
Database Solutions Sales Engineer
[Graphical user interface Description automatically generated with medium confidence]
clay.jackson_at_quest.com<mailto:clay.jackson_at_quest.com>
office 949-754-1203 mobile 425-802-9603
From: oracle-l-bounce_at_freelists.org <oracle-l-bounce_at_freelists.org> On Behalf Of Tharmarajah, Sam
Sent: Friday, October 22, 2021 11:53 AM
To: ORACLE-L <oracle-l_at_freelists.org>
Subject: Question: do you relink and bounce or just bounce or non, after your OS patching
CAUTION: This email originated from outside of the organization. Do not follow guidance, click links, or open attachments unless you recognize the sender and know the content is safe.
Hi All,
We are interested to know, what does the oracle community do when it comes to databases and os patching, especially when there is no kernel patch? The reason is that when there is no kernel patch, most of the times, our database servers doesn't get restarted. That would leave the databases up and running, without a restart. When reading metalink document, Doc ID 1467060.1 "Relinking Oracle HOME FAQ" , oracle recommends manually relinking the OH, after OS patching. Even though most of the times these OS patches had not caused any issue, there are occasions, where some databases have had performance degradations on these scenarios. Those times, just a restart addressed the problems.
To me, at least a restart of the databases after the OS patching will help to reinitialize the process with the new or updated libraries that are dynamically linked. Though I would prefer to relink and restart. But the systems folks think it could be an over kill, as most of the OS patches are quiet stable.
