RE: training for new DBA's
Date: Tue, 11 Feb 2014 10:13:53 -0700
Message-ID: <52FA5A51.3080308_at_1001111.com>
Communication is a wonderful thing. Too bad I am so poor at it :(
> From: Mark Bobak <Mark.Bobak_at_proquest.com>
> Date: Mon, 10 Feb 2014 19:47:05 +0000
> ³Back up the redo logs²
The database in question has a corrupt data file, other than that it is complete and is sitting on disk. When I say backup the online redo, I mean copy the redo and datafiles as Hans has mentioned. If a recovery goes bad:
How do you return to where you started if you do not have a copy of where you started.
Second, anyone can learn skills and technique, however, we have found that you cannot teach the correct attitude/mindset/paranoia. Either the candidate has it or they do not.
> Date: Mon, 10 Feb 2014 11:09:57 -0600
> From: Chris Taylor <christopherdtaylor1994_at_gmail.com>
>
> For this test that always fails, do you specify the Recovery Time Objective
> (RTO) - is it clear to these juniors what "success" is? I'm assuming it is
Success is not defined by recovery. It is defined by whether you took a copy of the corrupt database before you started or not. The candidate DOES NOT know this. The recovery is a difficult one. When I first ran into it 12 years ago it took 4 attempts to get it right.
> well defined, but I have a hard time with tests that are setup to fail.
> Juniors are going to make mistakes - that's why they are juniors and not
> intermediaries or seniors. Do they have access to a senior level during
> this test to gauge their input and is that input freely given?
They have had all the support they want or need for 3 months. If they do not have the correct attitude they are of no use to us.
Remember I am not testing skills or abilities.
> I could see sacking a junior for not engaging with team members to make sure they had
> their bases covered, but not for solely failing to backup redo logs in a
So, do I place a candidate who does not follow procedures in a client environment? They are not juniors yet, they are candidates.
> final test that even some intermediaries (and maybe a few seniors) would
> fail.
Our definitions of intermediates and seniors are different then.
(Granted, I'm probably reading more into your comment than intended
> and I sincerely apologize if so - I'm just genuinely puzzled by what you
> posted and I'm sure there's more to it than what you provided - at least I
> hope so)
See first line :)
> Date: Mon, 10 Feb 2014 14:56:14 -0700
> From: Hans Forbrich <fuzzy.graybeard_at_gmail.com>
>
> a) Never Back up the Redo Logs, and
> b) Before starting a recovery, they will take a safety COPY (but not
> back up) of the existing Redo Logs so they can restart a recovery if
> they botch the steps.
>
Amen
<tease>
Can you do that in OEM :)
</tease>
> I then go on to highlight some cases, and emphasize that more DBAs have
> been fired for violating these two points than any other reason (of
> which I am aware). 9 Well, they were not fired for violating the points,
> but for getting the recovery into an unmanageable and unrecoverable
> state, and having no plan B since they overwrote their Redo Logs with
> old copies.
Exactly
> With other teams "knowing better than the DBA" about what backups are
> required thus avoiding RMAN, and doing things like BCV splits and
> shoosting the mirror off to tape, it is highly likely that a) gets
> violated at some point.
Actually, if the two teams co-operate the use of silvering and mirroring (aka BCV, SRDF, etc) is just as safe. IF YOU HAVE THE CORRECT LEVEL OF PARANOIA!
>
> A Safety Copy in b) is tucked away in the DBA's home directory or some
> other non-traditional place;
> That Safety Copy should not be used, or required, unless the world has
> gone TU
Amen
John Hallas wrote:
> If you are going through that number of trainees and your success rate is only 40% then I seriously wonder how good your training is.
> You are investing 3 months of effort and then throwing 60% of it away.
> Perhaps it is better to look at your methods and trainers than just saying "Bye, Bye"
Actually what we need is a way to determine if the candidate has the correct attitude in the first week. That would improve our ROI immensely.
Donald Freeman wrote:
> I don't know. If you have very high standards and very serious consequences (think "Top Gun" or "Fukushima") then I can see washing out 60%.
> If your business model supports that then fine. It doesn't take a very big incident to cost millions. I don't know anybody personally whose salary would offset a big mistake.
The loss of a client's database would pretty well ruin our consultancy. No-one would trust us. Apart from the fact that placing a DBA without the correct attitude could open us to a charge of incompetence and/or negligence.
Apart form the fact how do I sit in front of a 15 year client and tell them "Although you only pay us do two things we are incapable of doing that" The 2 things being protect the data from loss and provide access to the data.
> Fair shout Donald and I don't want to get too bogged down into the original poster's situation.
> However a 60% failure rate does seem high after 3 months training. How many on this list, who consider themselves
> good/competent/reasonable DBAs would fail the same test. Is it the test , the training, the selection process? I don't know but it seems a flawed process to me.
Every good/competant/reasonable DBA could do the recovery. However, I doubt anyone, no matter their skill, would be successful the first time, unless you got very lucky. However, as mentioned, the recovery is not what you are judged on!
HTH
Dave
-- Dave Morgan Senior Consultant, 1001111 Alberta Limited dave.morgan_at_1001111.com 403 399 2442 -- http://www.freelists.org/webpage/oracle-lReceived on Tue Feb 11 2014 - 18:13:53 CET