| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 136.1 | Outline please? | MELKOR::SWIERKOWSKIS | | Thu Apr 16 1992 14:17 | 15 |
| Wendy,
Do you have a preliminary outline of the material or are you planning to
just lift the information straight out of the SYSNET curriculum for the
two-day course? Do you plan to just pull the remaining info from the
current VAXcluster Mgr course?
I have grave concerns if the above is true because the information will be
presented in the wrong order. We need an overview of the hardware (CI, DSSI,
Ethernet) and the SYSGEN parameters before we cover CLUSTER_CONFIG.COM.
A complete overhaul of the SYSNET course flow and this curriculum will be
necessary for this to be successful!
Susan
|
| 136.2 | cluster information | SUPER::WTHOMAS | | Thu Apr 16 1992 14:50 | 36 |
|
The preliminary course outline will be posted early next week.
(I'm waiting on this because I want the developers to have complete
buy-in before we commit to anything).
The two day course is to contain all of the Cluster information
that currently appears in the SYSNET curriculum (and yes, I am
following the discussion of cluster materials in those courses)
The challenge is to take that information and make it
instructionaly sound *without* adding new information (that the SYSNET
people did not have a chance to see).
Once the SYMAN people have taken the two day course, they are now
considered to be at the same level as the SYSNET people and they all
can now take the three day course which will cover the rest of the
cluster information (that was originally in the five day course).
The developers and I are going over each chapter and are really
working on making the information flow. I am also reviewing this course
from an Instructional Design perspective and am making some
changes/enhancements based on that.
The first chapter will be posted either Friday April 16th or the
following Monday.
Honest and constructive feedback is encouraged. Please keep the
feedback relevant to the course materials.
As we have had a few problems in the past, I feel it necessary to
say that any personal attacks on anyone involved in this curriculum
will be set hidden and the noter and the noter's manager will be
notified of this action.
Wendy
|
| 136.3 | FIX the SYSNET stuff FIRST!! | SOAEDS::TRAYSER | Seniority means a bigger shovel! | Thu Apr 16 1992 17:19 | 21 |
| Hold it! Time out! Is someone ignoring our previous input here?
The material on VAXcluster presented in the current curriculum is
INADEQUATE and poorly organized! I'm struggling through my second teach
of SysNet3 and have seen (and posted) numerous problems with this
material. THIS MATERIAL NEEDS TO BE FIXED FIRST! As Susan said, we need
configuration issues covered BEFORE we do CLUSTER_CONFIG! We need a good
discussion of Allocation class before we do CLUSTER_CONFIG!! We need an
understanding of MSCP servers before we can answer whether or not we are
going to server RF or HSC disks!!!
In other words, the current implementation has us telling people what to
type, but not WHY to type. No background for them to make a decision. I
have had NUMEROUS configuration questions from the SysNet3 students this
week (as well as the 1st time I taught it) and it is evident that THEY
think it is backwards.
PLEASE, get SysNet 2 & 3 fixed FIRST! Then, once they are successful,
modify the VAXcluster course.
$
|
| 136.4 | Okay, one more time.... | MELKOR::SWIERKOWSKIS | | Thu Apr 16 1992 17:47 | 51 |
| > -< FIX the SYSNET stuff FIRST!! >-
>
> Hold it! Time out! Is someone ignoring our previous input here?
Yes, I believe that to be the case! And it is frustrating!?
> The material on VAXcluster presented in the current curriculum is
> INADEQUATE and poorly organized! I'm struggling through my second teach
> of SysNet3 and have seen (and posted) numerous problems with this
> material. THIS MATERIAL NEEDS TO BE FIXED FIRST!
Two of my students stayed in the classroom for about ten minutes into the
lunch hour. Why?
1. To tell me that although I'm doing a great job, they have no choice but to
go home and tell their manager not to bother sending anyone to the new courses.
Why don't we in Ed Srv have something along the lines of a VMS Partner (someone
who is chartered with getting feedback like the above to engineering to get it
FIXED)?
2. To tell me that the only reason this course is working for them is because
I am not sticking to the material for either lecture or lab!!! If I were, they
would fail it/me!!!!!!!!
> In other words, the current implementation has us telling people what to
> type, but not WHY to type. No background for them to make a decision. I
> have had NUMEROUS configuration questions from the SysNet3 students this
> week (as well as the 1st time I taught it) and it is evident that THEY
> think it is backwards.
We absolutely have to get VAXcluster material in the correst sequence. Those
of us with a special love for clusters are painfully aware of what a major
disaster a minor misunderstanding regarding clusters can lead to. With DECnet,
even the most serious problems can be fixed with a rerun of NETCONFIG. The
more minor problems can usually be fixed with a few SET/DEF commands or
a simple shutdown and reboot. However, VAXcluster errors do things like
destroy the system disk, data disks, MSCP server, etc. Problems that
result in destroyed data and lots of downtime.
If I can guess the material that the two-day VAXcluster class will contain
by the titles of the modules in this conference, it is NOT material that is
adequately covered in SN 2 & 3. Of course that means that the students who
take the three-day follow on class from SYSNET won't have the same background
as those from SYSMAN. We're creating another problem unnecessarily.
> PLEASE, get SysNet 2 & 3 fixed FIRST! Then, once they are successful,
> modify the VAXcluster course.
YES, YES, YES!!!!
Susan
|
| 136.5 | First things first!!! | SOAEDS::HILLMAN | Network partner excited..! | Fri Apr 17 1992 09:51 | 8 |
| I agree with [.3]; since the first teaches for each SYSNET II & SYSNET
III, I have not followed the course outline (bubble chart) because it
doesn't seem to flow logically... SYSNET III could very well do without
Module 1 in my opinion. Invariably, the technical depth of the material
"teases" the students and they ALWAYS ask questions of a deeper nature
that are sometimes covered in later chapters, so it's extremely
difficult to maintain the flow according to the book.
|
| 136.6 | Status of VAXcluster System Management ? | NEURON::STAHLY | 10$: BRB 10$ | Fri Apr 17 1992 11:20 | 9 |
|
What then is the status of VAXcluster System Management (5 days) ?
Will it be updated before this material is 'pulled' to create the
3-day course, or will it remain outdated ?
Ron
|
| 136.7 | VAXclsuter not SYSNET | HARDY::MOSTEIKA | Paul, ZKO1-1/D42, 381 (881)-1075 | Fri Apr 17 1992 13:58 | 37 |
| No we are not ignoring previous requests for fixing the material. A lot of
fixes were implemented as the TBI version of the SYSNET courses were created. If
we source that text, and make other corrections as we go, sequencing etc., then
we should be all set.
One dilemma is that the two day must cover the SYSNET cluster info missed by the
SYSMAN folks and nothing more. We can fix, resequence, modify and such, but we
can't really add without slighting the SYSNET people. The other catch is the
NETWORK stuff required for a cluster build probably won't be included. We can
possibly have NETMAN I as a prereq..
As far as configuration issues, I agree. But how much? More than CI, NI, mixed
possibly a DSSI and FDDI? What else?
MSCP and the params (MSCP_LOAD...), allocation class, disk servers like you said
are covered in SYSNET II (Chap 1). Maybe we do need some review in SYSNET III
before cluster config, but what do we steal from to do it? Performance? It's not
really fair to those that do abide by the suggested prereq's.. In the pilot of
SYSNET II, where performance originally was, many students were interested in
the performance piece.
I know it must be hard to teach tasks or cookbook style without the theory.
Historically we never really did it that way. However, if you add too many
"why-fors", you have the problem of not covering all the topics. The objective
of SYSNET was to include management of VAXclusters and Networked systems in a
task oriented manner. Pieces from six different courses were rolled into three.
There's no time to cover all the theory, as you are experiencing.
SYSNET was our first attempt at a pretty radical feat. While not perfect, we did
have some good comments, and some good classes as a result. With your help, we
can make SYSNET and VAXcluster even better. However, right now we have to focus
on the VAXcluster - that's what they are paying us to do.
Regards,
Paul M.
|
| 136.8 | Update | HARDY::MOSTEIKA | Paul, ZKO1-1/D42, 381 (881)-1075 | Fri Apr 17 1992 14:13 | 9 |
| > Will it be updated before this material is 'pulled' to create the
> 3-day course, or will it remain outdated ?
We will update. Any specific problem areas or suggestions, please let us know.
Regards,
Paul M.
|
| 136.9 | Update VAXcluster System Management ? | NEURON::STAHLY | 10$: BRB 10$ | Fri Apr 17 1992 14:35 | 13 |
|
>>>We will update. Any specific problem areas or suggestions, please
>>>let us know.
Are you really going to update the existing VAXcluster System Management
Course ? I need to know the answer to this question, course schedules
need to be submitted. What is the time frame for completion ?
Yes I will provide input.
I will keep this a separate reply. Please respond to it.
Ron
|
| 136.10 | We CAN'T assume they take SN2 (much less remember it all)!!!! | SOAEDS::TRAYSER | Seniority means a bigger shovel! | Sat Apr 18 1992 04:38 | 95 |
| |The other catch is the
|NETWORK stuff required for a cluster build probably won't be included. We can
|possibly have NETMAN I as a prereq..
I can live with this. I've always suggested to my students that Network
I be taken before VAXcluster if they have the time and money for two
courses.
|As far as configuration issues, I agree. But how much? More than CI, NI, mixed
|possibly a DSSI and FDDI? What else?
Bingo! You've got the problem! VAXclusters is too complex of a topic to
believe that you can give a cursory introduction to it and NOT expect the
students to ask questions. This week I had CI, NI and MI students. The
information I gave them just triggered a couple dozen questions. We need
to examine this carefully. "Trust me. Just climb in the seat, press the
clutch, turn the key, put it in 1st, slowly release the clutch and...
VOILA! You are now driving!"
|MSCP and the params (MSCP_LOAD...), allocation class, disk servers like you said
|are covered in SYSNET II (Chap 1). Maybe we do need some review in SYSNET III
Right. Information like this needs to be refreshed before doing a
configuration.
|before cluster config, but what do we steal from to do it? Performance? It's not
|really fair to those that do abide by the suggested prereq's..
"It's not really fair to those that do abide by the suggested prereq's.. "
Ah Ha!! That is a major part of the problem. The people that took Sys
Mgr II didn't necessarily take Sys Mgr I, and if they did it might have
been months previous. A majority of my Sys Net 3 students have NOT taken
Sys Net 2!!!!! Somewhere up in the corporate marketing dungeons the
people wandering the halls have forgotten that many students don't take
the SPECIFIC prerequisite course before attending these courses! Look at
page XXV:
NOTE
These prerequisites can be satisfied by taking the VMS System and
Network Management II: Managing Established Systems course OR MAY BE
OBTAINED THROUGH EXPERIENCE operating or managing a VMS system.
We have just told them "you don't really have to take Sys Net 2, just be
experienced"! Fine. Now, we turn right around and say, "Sorry, the
parameter ALLOCLASS was covered in Sys Net 2...I can't explain it to you
in this class." No, *NOTHING* on the PREREQUISITE page states ANYTHING
about having a VAXcluster background information before attending SN3!!
We are ASSUMING that the student will 1) take SN2 and 2) remember obscure
SN2 topics.
|In the pilot of
|SYSNET II, where performance originally was, many students were interested in
|the performance piece.
Its an in-depth topic and doesn't belong in SN2, and although it would
fit in SN3, there isn't time -- VAXclusters and Networks are more
important.
|I know it must be hard to teach tasks or cookbook style without the theory.
|Historically we never really did it that way. However, if you add too many
|"why-fors", you have the problem of not covering all the topics.
True. However, by the 3rd or 4th course the students often WANT the
theory -- helps them understand well enough to make their own decisions
regarding installations and configurations. If there are too many
topics, move them to another course!
|There's no time to cover all the theory, as you are experiencing.
Right. Problem is we don't have a class full of operators by SN3! We
have a lot of technical professionals -- people that don't want STRICTLY
a cookbook.
|However, right now we have to focus on the VAXcluster - that's what they are
paying us to do.
AARRGGHH! How can we POSSIBLY consider chopping up another course that
technically follows SN3 until we decide how to fix SN3!? If we do this
then we are basically saying SN3 stays as it is. This is a mistake. We
need to at least iron out what goes in SN3 so we can better decide what
needs to be in the new VAXcluster course.
|SYSNET was our first attempt at a pretty radical feat. While not perfect, we did
|have some good comments, and some good classes as a result. With your help, we
|can make SYSNET and VAXcluster even better.
"With your help..." OK, I hope you mean this. I'm a tough critic. I'll
vent my opinions here with conviction. Of course I'm the "good corporate
player" while in class -- never a criticism in front of the students
regarding the material, but it has been real tough with SN3. We need to
look at it carefully considering our audience and our prerequisites and
our ASSUMPTIONS of what type of student will be attending it.
$
|
| 136.11 | Prereq's, A Little Review Goes a Long Way | HARDY::MOSTEIKA | Paul, ZKO1-1/D42, 381 (881)-1075 | Mon Apr 20 1992 11:14 | 23 |
| We could never enforce prereq's - even internally. That's why I gave up and
just taught them. Of course some argue that you will loose the rest - it usually
turns out they need the review. With any detailed subjects you have the problem
with time. Hold remedial training classes. Sounds easy but it's a lot of extra
work.
Let's face it, I doubt if we could ever create a course to satisfy all the
various prereq problems. If we did, the course would have to be two weeks long.
That won't happen. You know what that leaves; you're saying that there's not
enough time now. So the instructor must use their own judgement to re-emphasize
or review important information according to the experience level of the class.
Eqivalent experience means just that. We do list the course goals of the
required course in with the course description. Manage a one node VAXcluster
does leave it somewhat open.
"With your help..." - sure I mean it. Vent away.
Regards,
Paul M.
|
| 136.12 | Let's really make this work. | MELKOR::SWIERKOWSKIS | | Mon Apr 20 1992 21:49 | 49 |
| As I see it, the biggest problem facing us in this project will be to take
the material covered in SYSNET 2 and 3 and make it a meaningful two-day
course. To recap:
The SYSNET courses cover (VERY superficially) allocation classes,
NETCONFIG.COM, Quorum, Common System Disk directory structure,
CLUSTER_CONFIG.COM, SHOW CLUSTER Utility, SYSMAN and an incomplete
SYSTARTUP_V5.COM.
The rest of the topics in the current VAXcluster class are:
1. Overview of the hardware and software, including SCA.
2. Planning the VAXcluster.
3. More details on SYSTARTUP_V5.COM, SHOW CLUSTER and troubleshooting.
If we follow the current plan, we will be building a VAXcluster (in the
two-day class and SYSNET 2 & 3) *BEFORE* we understand what a VAXcluster
is (overview in the three-day, follow-on class).
We are trying to cram too much information too fast in the beginning, and
we don't have enough meat left over for the end.
> There's no time to cover all the theory, as you are experiencing.
The only way to satisfy the customers' need is to make the time to adequately
cover enough theory so they can make intelligent decisions. I warn them in
the beginning that I will be talking very fast to get it all in. Believe me,
they recognize the time constraints and write fast. I then have to skimp
on the performance topics, but by Friday most people are fried anyway. They
feel like they've accomplished something when they can build a network node
and an NI VAXcluster that work. These guys are technical professionals (not
Operators) as Buck said earlier. We are CHEATING them!
> However, right now we have to focus on the VAXcluster - that's what they are
> paying us to do.
This really concerns me. We're doing what "they" are forcing us to do
instead of doing what's right for the customer. How can we expect to come
up with quality products when we ignore the customer comments?
Like Buck has already stated, we have no business chopping up another
course until we fix the problems in the SYSNET curriculum (more than just
resequencing and fixing typos).
Susan
|
| 136.13 | working course outline | SUPER::WTHOMAS | | Tue Apr 21 1992 11:30 | 59 |
|
In response to the requests here and after looking at the SYSNET
cluster material ourselves and having some difficulty putting it all
together, we have decided to work from the original cluster management
materials for the two day course.
For the Two day course the chapters will be as follows:
1. Introduction to VAXcluster environment
The VAXcluster Concept
VAXcluster Components
Cluster Integrity
Connection Manager
2. Building a VAXcluster
Building a VAXcluster System
VAXcluster Management and Security
Selection of Special Purpose Members
VMS Installation
Installing Required Software License Keys
Configuring the DECnet VAX Network
Using Cluster_Config.com
Configuring System Disks
Startup Command Procedures
System Disk Structure
3. Managing VAXcluster Operations
Sysman Commands
LMF
System Time in a VAXcluster System
Updating VMS in a VAXcluster system
Gathering Diagnostic Information
Appendix - License Command Descriptions
Note that this is a working outline, the development team is going
to sit down and review *every* piece of information that will go into
this course to decide if it is at the appropriate level and if
additional information is needed in order for it to be understood by
the student.
We will also be updating the cluster information to conform to 5.5
standards.
As far as fixing the Sysnet information, if the information for the
is course is seen as an improvement, we could consider retrofitting the
information back to Sysnet but we will have to wait and see.
Wendy
|
| 136.14 | | AUBREY::DONHAM | Progress Through Tradition | Tue Apr 21 1992 12:36 | 30 |
|
Just poking my nose in...I'm the other developer working on this.
To recap, what we're doing is this:
1. Pulling the material common to SYSNET and CLUSTER from CLUSTER
2. Comparing the material to the current SYSNET for update information
3. Examining the material for possible updates to V5.5
4. Checking the material against NOTES feedback for possible modification
This information becomes the 2-day crossover course from SYSMAN to SYSNET. It
may sound backward to start with CLUSTER, then update it from SYSNET info,
but we found that pulling 'updated' info directly from SYSNET gave us a lot
of text that lacked context. Pulling it from CLUSTER preserves the context
and sequencing.
The next step is to do a similar operation for the material that appeared in
CLUSTER which didn't make it to SYSNET. This information will be updated and
expanded slightly to create the 3-day course that both SYSMAN and SYSNET students
will enter. There will be some redundancy in both streams, but it's necessary
to make certain that everyone is at the same starting point at the beginning
of the 3-day piece.
Oh, one more thing...Buck...don't look now, but there's a B100 parked
behind your cube! ;^)
Perry
|
| 136.15 | | HARDY::MOSTEIKA | Paul, ZKO1-1/D42, 381 (881)-1075 | Tue Apr 21 1992 13:03 | 30 |
| It wont be easy given the constraints, but I think we can make it meaningful.
That's what we need, focused attention on certain topics, deficiencies, (not
the old war of task vs. theoretical).
Did you really think that quorum and Cluster_Config.Com was too light? Take
quorum...
we could emphasize the system parameters a little more & give an example
of quorum during boot-formation, what else? Do they really need to know
all the particulars of the actual formula? We give them some of that in
text. The Instructor Notes allow some added value. What else?
We'll look at them more closely. As far as the sequence, building before
planning, we have the constraint of just the topics in SYSNET for the two day'r.
We give them some planning -security, etc. But not much. Yes the hardware and
software will come later. But if we give them what they need to do the mechanics
of building, then you can elaborate later with the theory and particulars.
In another sense, we always tend to teach them everthing they wanted to know and
more. By the time they got to utilize the theory, they forgot what they're doing,
and possibly even why. You can feed them the detail in different stages, adding
to the complexity. In fact it's preferred by many. Reinforcement by review.
As far as the rest... I'm only the piano player... Just curious, did you or $
participate in the Instructor Forum about the SYSNET Curriculum?
Regards,
Paul M
|
| 136.16 | Instructor Forum? What Instructor Forum? | SOAEDS::TRAYSER | Seniority means a bigger shovel! | Tue Apr 21 1992 13:54 | 0 |
| 136.17 | schedule estimate | SUPER::WTHOMAS | | Tue Apr 21 1992 14:17 | 15 |
|
The *rough* schedule for these courses is to get the two day course
out by mid May and the three day course out by the end of June.
We do plan on updating the current cluster information but it will
not immediately show up in the original cluster course. (I don't think
that our updates should affect anyone's schedules)
After that will come the plans for a one day seminar on MDF but
that is another story (and note).
Seeing as the course outlines were just reworked, the project plan
needs work and will be posted in just a few days.
Wendy
|
| 136.18 | Which instructor forum...
| SUPER::MATTHEWS | | Tue Apr 21 1992 14:31 | 4 |
| re .16 The one held on September 4-5, 1990 in Chelmsford MA. (The most
recent one.)
Val
|
| 136.19 | I'm sure somebody went... | SOAEDS::TRAYSER | Seniority means a bigger shovel! | Tue Apr 21 1992 19:19 | 17 |
| But if I recall correctly it was canceled, then
rescheduled with only a few weeks notice -- I couldn't
change my schedule.
Sigh! 1990! Our staff has had almost a 90% turnover
in the last two years!
About a year and a half ago...harumph. I don't think
I ever saw any minutes from that meeting posted anywhere.
I'm sure I looked over the outlines, but until some
flesh has been hung on the bones its hard to get a really
good review.
SO, when is the NEXT meeting??
$
|
| 136.20 | Why does this need debate? | NITTY::THORNE | Department of Redundancy Department | Thu Apr 23 1992 11:27 | 31 |
| Re: .15
Yes, I was at that Instructor Product Forum. I can tell you that the
implementation of this curriculum has strayed a bit from the original
design. It was stated that the big goal in going to the new approach
was that students could gain the types of skills necessary for most of
them to do their jobs by taking fewer courses. That's not really
happening. Nobody gets all the material they would have received by taking
the old curriculum courses by taking the new curriculum courses. I'm also
not convinced that the basic parameters given for the cluster class are
valid. I wonder at the course length, and the topics covered seem dilute.
I know, I know, that none of these issues were the developers' ideas.
It seems that we all have to live with this.
I agree with notes written here by Buck and Susan and others about
improvements that are necessary in the SYSNET curriculum. I am
gratified to see that the entreaties to change the approach to the
cluster course are seeing some light. I wish that it wasn't necessary
to go through all of this verbal fencing to get anything changed. It
seems to waste a good deal of everybodys' time -- and we don't really
have the time to waste. While we debate the various course issues
here, most of our competitors can develop a tailored course to a customer's
exact specifications (and probably deliver it as well). They simply
don't insist on the lengthy bureaucratic rituals that we do. They know
it and the customers know it. If the fixes to the content and flow of
this curriculum aren't performed centrally, then we'll probably do what we,
as instructors, have done in the past; augment the materials and labs and
rearrange topics to make it work and GET THIS DONE.
Mark Thorne
Chicago Training Center
|
| 136.21 | Still hoping.... | MELKOR::SWIERKOWSKIS | | Thu Apr 23 1992 15:12 | 87 |
| Paul (and the other Developers poking in....)
>Did you really think that quorum and Cluster_Config.Com was too light? Take
>quorum...
> we could emphasize the system parameters a little more & give an example
> of quorum during boot-formation, what else? Do they really need to know
> all the particulars of the actual formula? We give them some of that in
> text. The Instructor Notes allow some added value. What else?
Yes, please. Let's get this into the light of reality (ie what goes on in a
classroom full of technically competent professionals). Quorum is a 1 hour
lecture and has been that way ever since I started teaching VAXcluster under
V4.x. The issue here for system managers is not limited to setting the SYSGEN
parameters for VOTES and EXPECTED_VOTES. They absolutely must understand the
impact on Quorum of booting nodes into and removing nodes from the VAXcluster.
When does the VAXcluster hang and why? What can they do to make their cluster
more available (Quorum Disk)? What can they do in emergencies (too many
systems crashing/shutting down)? Do they have to halt and reboot the entire
cluster (conversationally, to change EXPECTED_VOTES) or can they try to adjust
Quorum at the console (>>>D/I 14 C). Will $SET VOTES/EXPECTED help them or
leave them in the same predicament they were in before they executed the
command? All of this is REQUIRED information regardless of which side of the
theory v. cookbook fence you're on. And, of course, the value of VOTES on
satellites adds another angle to the whole discussion.
Even in the 38400 baud mode of lecture I've had to do in the new courses,
students continue to tell me I have finally solved their problem(s) for them.
In other words, they now know how to make the cluster, network, whatever, do
what they NEED!!!!
>We'll look at them more closely. As far as the sequence, building before
>planning, we have the constraint of just the topics in SYSNET for the two day'r.
>We give them some planning -security, etc. But not much. Yes the hardware and
>software will come later. But if we give them what they need to do the mechanics
>of building, then you can elaborate later with the theory and particulars.
Hasn't the Operator class already experienced problems with the hardware/
software overview at the end? I haven't taught that course (and probably
never will given my current schedule), but the comments in the Operator Note
have led me to believe that most instructors are changing the order and doing
those topics first. As many of us have already stated in one form or another,
how are you supposed to get through the questions in CLUSTER_CONFIG.COM on
MSCP, RF disks, HSC disks, NISCS_xxxx without an overview first?
>In another sense, we always tend to teach them everthing they wanted to know and
>more. By the time they got to utilize the theory, they forgot what they're doing,
>and possibly even why. You can feed them the detail in different stages, adding
>to the complexity. In fact it's preferred by many. Reinforcement by review.
Well, our experiences don't support that. My VAXcluster System Manager
students can ALWAYS build a cluster on Wednesday afternoon after being
"inundated" with three days of theory. (Ditto, my Network Managers in
Network Management I). However, my SYSNET III students can NOT do it without
a hand-holding cookbook type of lab because one day doesn't give them what
they need to be successful. (As I've already promised Buck, I will post
those labs in the appropriate Note.) The students who have trouble with the
amount of theory we taught in the past are the ones who attempt to enter DCL
commands at an * prompt and insist something's wrong with our system. In
other words, they need a cookbook and one-on-one training. They don't have
the skills/background to survive in a technical classroom. The new curriculum
doesn't help them either; in fact, it makes it worse because there is so
little time for review.
>As far as the rest... I'm only the piano player... Just curious, did you or $
>participate in the Instructor Forum about the SYSNET Curriculum?
I understand that you don't have the freedom/power to really address these
issues, but I'm hoping you can get the attention of someone who can. About
that Forum? HAHAHAHAHAHAHAHAHA! Whew! That was really funny! You do
understand that most Instructors do not have the opportunity to participate
in such activities (and as Mark mentioned, it doesn't seem to make much
difference anyway)? As far as Santa Clara's ability to participate, we
managed to get one instructor to the pilots in LA; too LATE to make much
difference. We were involved in a major remodel of our training center at
the time. We moved into temporary quarters for four months and then moved
back again and merged with Customer Services Training. We have also lost
half of our senior staff in four months. Just getting through Monday has
been a challenge for us.
One more time (broken record here), unless we can get a FUNDED review cycle
in the field, our problems will never be resolved. I'm currently scheduled
to teach eleven weeks this quarter. I'm trying to learn Ultrix, Alpha,
and NT. AND I'm working on my own labs for SN 2 & 3. AND I'm trying to stay
current in this conference. AND I'm trying to find time for my family! And
Paul, thanks for listening.
Susan
|
| 136.22 | Once more unto the breach! | SONATA::SADLER | Change for a Flainian Pobble Bead? | Tue Apr 28 1992 14:51 | 43 |
| Re: .4
>Two of my students stayed in the classroom for about ten minutes into the
>lunch hour. Why?
>
>1. To tell me that although I'm doing a great job, they have no choice but to
>go home and tell their manager not to bother sending anyone to the new courses.
Susan, I'd appreciate your mailing me details of these 2 students. I'd
like to get a first-hand follow-up on this to understand why they feel
this way. Thanks.
Re: .21
>One more time (broken record here), unless we can get a FUNDED review
>cycle in the field, our problems will never be resolved.
I did fund field review for the Sysnet courses, in fact the funded
field reviewers taught the pilots!
Re: The whole discussion...
We're meeting this afternoon to try to resolve all the issues around
VAXclusters, Sysnet I-III and also the Network content. We'll come up
with a recommendation which we'll then post here for review. In
addition, to ensure that we get real agreement from all training
centres, we'll be sending it out to the 3 Geographies via the
designated contacts (Sherry Butler,Nancy Giddens, Bob Sowton).
We'll ensure that any necessary fixes to Sysnet are made either before
or in parallel to the VAXcluster rework.
Watch this space.
Andy
|
| 136.23 | What "funded" review? | MELKOR::SWIERKOWSKIS | | Tue Apr 28 1992 20:57 | 21 |
| > I did fund field review for the Sysnet courses, in fact the funded
> field reviewers taught the pilots!
Andy,
Who did what and when? Would it have been possible to completely overhaul
the curriculum when the pilots were in November and delivery was scheduled
for January? I don't know any reviewers, and to the best of my knowledge,
no one in Santa Clara has been a reviewer in the last half a dozen years.
Several years ago I was the designated reviewer for the rewrite of the
VAXcluster Operator course. My schedule was cleared for two weeks, but the
course never materialized.
As for being able to review modules as they become available, most of us are
either out of town or swamped with a heavy schedule or buried in another
project. I've also noticed that by the time the modules are up for review,
it's too late to make anything but superficial fixes (typos) before final
printing. And, we've already gone around and around on what it takes to make
"quick" fixes to our material.
Susan
|
| 136.24 | VAXcluster Problem Report -CSC/Questionnaire Addressing Service Cost Reduction | HARDY::MOSTEIKA | Paul, ZKO1-1/D42, 381 (881)-1075 | Wed Apr 29 1992 11:00 | 1063 |
| The following is a report catagorizing over 700 VAXcluster calls at the CSC
and soliciting your input to help reduce Service Cost. They want to hear the
opinions from the experienced personnel.
There's a node listed for access to the PostScript file (and this .TXT file)
Regards,
Paul M.
VAXclusters
VAXclusters
VAXclusters
Services Cost Reduction Questionnaire
Services Cost Reduction Questionnaire
Services Cost Reduction Questionnaire
03-Apr-1992
Abstract
Abstract
Abstract
This questionaire is part of an on-going effort by the Contin-
uing Improvement Engineering Group within Digital Services to
reduce the costs of providing support for VAXclusters.
The purpose of this questionaire is to solicit recommendations
for changes to specific functions and components of the VAXclus-
ter Architecture.
Digital Equipment Corporation
Digital Equipment Corporation
Digital Equipment Corporation
Maynard, Massachusetts
Maynard, Massachusetts
Maynard, Massachusetts
1 Project Background
1 Project Background
1 Project Background
The project effort to date has consisted of analyzing and cate-
gorizing VAXcluster calls logged during a six week period in the
Customer Support Center/Colorado call database.
Over 700 calls have been analyzed, and grouped into maajor
categories:
Requests for Infor- 48.6%
mation
Severe System Dis- 29.5%
ruption
System Failure 21.9%
These groupings include repeat and recurring calls. They form
the basis for the questions in this document.
NOTE
NOTE
NOTE
______ __ ________ ___ __ ____ __________ ___ _____ ___
Issues of training and on site consulting are being ad-
Issues of training and on site consulting are being ad-
Issues of training and on site consulting are being ad-
_____ __ _____ ______ ______ ________
dress by other groups within Digital.
dress by other groups within Digital.
dress by other groups within Digital.
2 Guidelines for Completing This Questionnaire
2 Guidelines for Completing This Questionnaire
2 Guidelines for Completing This Questionnaire
MAY 15, 1992
MAY 15, 1992
MAY 15, 1992
o Responses are needed by .
o You do not have to respond to every question. The intent of
this questionnaire is to gather "opinions" based on "real-
world experience."
o Please limit your responses to the questions asked. It is
beyond the scope of this questionnaire to provide a forum for
a general discussion of cluster issues.
Digital Equipment Corporation - Company Confidential 1
Digital Equipment Corporation - Company Confidential 1
Digital Equipment Corporation - Company Confidential 1
o Read the next section titled
______ __ ___ _____________
Format of the Questionnaire
Format of the Questionnaire
Format of the Questionnaire
before you answer any of the questions. It explains how
questions are grouped and how "introductory" text precedes
each grouping.
o Send your completed questionnaire via Inter-Office Mail
(paper-mail) to:
Harold Owens
Harold Owens
Harold Owens
ZKO1-1/F22
ZKO1-1/F22
ZKO1-1/F22
o Additional copies of this questionnaire can be copied over
the E-net from:
- CSSE32::SYS$PUBLIC:CLUSTERS_QUESTIONAIRE.PS
- CSSE32::SYS$PUBLIC:CLUSTERS_QUESTIONAIRE.TXT
2 Digital Equipment Corporation - Company Confidential
2 Digital Equipment Corporation - Company Confidential
2 Digital Equipment Corporation - Company Confidential
3 Format of the Questionnaire
3 Format of the Questionnaire
3 Format of the Questionnaire
Calls have been grouped into major categories based on the type
of operation, or cluster component affected by the customer's
actions:
Configuration 80% of all calls
Communications 15% of all calls
Storage 5% of all calls
A summary of the issues contained in a each category is pre-
sented first. The summary is intended to give you a "feel" for
the issues raised. You should then have a proper context in
which to respond.
Each major category is further divided by VAXcluster feature or
function:
Digital Equipment Corporation - Company Confidential 3
Digital Equipment Corporation - Company Confidential 3
Digital Equipment Corporation - Company Confidential 3
CONFIGURATION
CONFIGURATION
CONFIGURATION
<- major category
- \
- - summary of issues
- /
Modify, Add, Delete...
Modify, Add, Delete...
Modify, Add, Delete...
- \
- - call examples
- /
...
(next subcategory)
(call examples)
...
CONFIGURATION QUESTIONS
CONFIGURATION QUESTIONS
CONFIGURATION QUESTIONS
-
-
-
Questions and space for your responses are presented last.
4 Digital Equipment Corporation - Company Confidential
4 Digital Equipment Corporation - Company Confidential
4 Digital Equipment Corporation - Company Confidential
4 CONFIGURATION
4 CONFIGURATION
4 CONFIGURATION
Installing VMS "out of the box" and selecting "VAXCLUSTER" as an
install option goes rather smoothly. Making the transition from
one configuration to another, or modifying an existing configu-
ration requires a bit more effort, and a lot more knowledge.
The following call samples highlight the types of problems
categorized as "CONFIGURATION."
NOTE
NOTE
NOTE
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
" ) are random calls and solutions taken from the CSC/CO
" ) are random calls and solutions taken from the CSC/CO
" ) are random calls and solutions taken from the CSC/CO
call database
call database
call database
4.1 Customer Questions Include...
4.1 Customer Questions Include...
4.1 Customer Questions Include...
Modify, Add, Delete...
Modify, Add, Delete...
Modify, Add, Delete...
- How to add a device (node, disk, hsc, etc.) to a cluster?
- How to delete a device (node, disk, hsc, etc.) from a
cluster?
- How to modify component designations (node name, group
number, allocation class, etc.)
- "Removed a satellite node from a cluster and made it a
standalone cluster. The old boot node is logging errors on
pea0 and errlog shows invalid cluster password received...
SOLUTION: The cluster group # was the same on the new
system as the old system. Changed cluster group # on new
system and stopped entry on pea0..."
Digital Equipment Corporation - Company Confidential 5
Digital Equipment Corporation - Company Confidential 5
Digital Equipment Corporation - Company Confidential 5
- "Customer getting CSP error on $GETDVI when systems trying
to access quorum disk.
SOLUTION: Customer had DISK_QUORUM set to wrong HSC name.
Customer changed it to correct HSC node name..."
Cluster Parameters...
Cluster Parameters...
Cluster Parameters...
- What changes are required to SYSGEN parameters to ADD
/DELETE/MODIFY nodes?
- What changes are required to MODPARAMS.DAT to ADD/DELETE
/MODIFY nodes?
- What parameters need adjusting after AUTOGEN runs?
- "Customer wants to merge two 3100 satellites on a CI
cluster. What is BOOTNODE parameter in MODPARAMS.DAT?
SOLUTION: BOOTNODE is used by AUTOGEN to set LOCKDIRWT to
'1' when it equals 'TRUE.' Also used to indicate that the
node is a BOOT server..."
- "2 3400's hang when booting and forming a cluster.
SOLUTION: EXPECTED-VOTES incorrect."
- "Errlog filling up ... getting invalid cluster password.
SOLUTION: Set VAXCLUSTER to 0 until the additional CI node
arrives."
Interconnects...
Interconnects...
Interconnects...
- "Question on cables for CI ... Can I power down a node
and disconnect its CI cable and move them to a new Star
Coupler?
SOLUTION: Yes with no problem; the star is a passive
device"
6 Digital Equipment Corporation - Company Confidential
6 Digital Equipment Corporation - Company Confidential
6 Digital Equipment Corporation - Company Confidential
- "Brian has a DH 4000 and a 3800 connected with DSSI. What
is happening is when he unplugs the Ethernet cable from
the 3800 side, it crashes
SOLUTION: Spoke with Kevin about this. Extending RECNXTIN-
TERVAL to a few minutes should take care of this."
Quorum...
Quorum...
Quorum...
- What is quorum?
- How to set up quorum?
- What role does quorum play during a Cluster State Transi-
tion?
- What is a quorum disk?
- When do I need a quorum disk?
- How to set up and use a quorum disk?
- What is a quorum watcher?
- "Erick has a 3 node CI cluster and wants to set up a
quorum disk so that 1 node can run by itself.
SOLUTION: Define the DISK_QUORUM value in SYSGEN to be
the same for all nodes, and set QDSKVOTES to 2. This will
establish a quroum of 3."
Digital Equipment Corporation - Company Confidential 7
Digital Equipment Corporation - Company Confidential 7
Digital Equipment Corporation - Company Confidential 7
- "Ali has a three node CI cluster where quorum = 2 and
total votes =3. He needs to take one machine down for
upgrade. His questions: (1) Can he take one machine down
without disrupting the other two? (2) Does he need a
quorum disk?
SOLUTION:(1) Yes, however there will be a short cluster
state transition. (2)No, the two machines will continue to
function. However ... "
- "Two days ago, a message appeared on the console stating
that quorum had been lost. It eventually was regained but
Bill was some what upset because he issued a SHOW CLUSTER
and saw that his HSC's were not members.
SOLUTION: Customer was not too familiar with cluster
concepts. Explained that the HSC's would be seen but
wouldn't be members of the cluster. What he sees is OK.
Checked values of QDSKINTERVAL and this was set to 20 on
both nodes. He doesn't recall any unusual activity when
this happened."
8 Digital Equipment Corporation - Company Confidential
8 Digital Equipment Corporation - Company Confidential
8 Digital Equipment Corporation - Company Confidential
5 CONFIGURATION QUESTIONS
5 CONFIGURATION QUESTIONS
5 CONFIGURATION QUESTIONS
5.1 Modify, Add, Delete...
5.1 Modify, Add, Delete...
5.1 Modify, Add, Delete...
1.
1.
1.
VAXcluster management requires a lot of "user intervention."
What changes to the user interface would you recommend to
simplify the management process?
2.
2.
2.
When other people use you as an information source, what
information/advice do you most often pass on in regards to
VAXcluster management?
Digital Equipment Corporation - Company Confidential 9
Digital Equipment Corporation - Company Confidential 9
Digital Equipment Corporation - Company Confidential 9
5.2 Cluster Parameters...
5.2 Cluster Parameters...
5.2 Cluster Parameters...
1.
1.
1.
The various utilities and procedures that comprise the "tool
set" for VAXcluster creation and control - SYSGEN, SYS-
MAN, CLUSTER_CONFIG.COM, AUTOGEN, etc. - present a number
of choices which a Cluster Manager must make. What recom-
mendations would you make to reduce the number of controls
required for VAXclusters?
10 Digital Equipment Corporation - Company Confidential
10 Digital Equipment Corporation - Company Confidential
10 Digital Equipment Corporation - Company Confidential
2.
2.
2.
When other people use you as an information source, what in-
formation/advice is most often asked of you about VAXcluster
parameters?
Digital Equipment Corporation - Company Confidential 11
Digital Equipment Corporation - Company Confidential 11
Digital Equipment Corporation - Company Confidential 11
5.3 Interconnects...
5.3 Interconnects...
5.3 Interconnects...
1.
1.
1.
The physical interconnect allows the VAXcluster software
to communicate across all nodes in the cluster. The three
physical links - Ethernet, CI, and DSSI - have unique char-
acteristics, and unique management issues that must be dealt
with. When other people use you as an information source,
what information/advice do most often give regarding inter-
connects?
5.4 Quorum...
5.4 Quorum...
5.4 Quorum...
1.
1.
1.
Issues surrounding quorum - definition, usage, etc. - con-
stitute a large large portion of the calls analyzed. What
changes would you like to see implemented in the VAXclus-
ter architecture to improve the implementation of the quorum
scheme?
12 Digital Equipment Corporation - Company Confidential
12 Digital Equipment Corporation - Company Confidential
12 Digital Equipment Corporation - Company Confidential
2.
2.
2.
When other people use you as an information source, what
information/advice do you most often give regarding quorum?
Digital Equipment Corporation - Company Confidential 13
Digital Equipment Corporation - Company Confidential 13
Digital Equipment Corporation - Company Confidential 13
6 COMMUNICATIONS
6 COMMUNICATIONS
6 COMMUNICATIONS
Internode communications and network access, requires an under-
standing of protocols, databases, utilities, and commands. All
of this adds to the complexity of managing a VAXcluster.
The following call samples highlight the types of problems
grouped in the "COMMUNICATIONS" category.
NOTE
NOTE
NOTE
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
" ) are random calls and solutions taken from the CSC/CO
" ) are random calls and solutions taken from the CSC/CO
" ) are random calls and solutions taken from the CSC/CO
call database
call database
call database
6.1 Customer Questions Include...
6.1 Customer Questions Include...
6.1 Customer Questions Include...
- How to define and use internode access?
- How to define and use network access?
- How to define and use MOP and BOOT servers?
- "Customer ran CLUSTER_CONFIG.COM to convert cluster from
a CI to an MI. After he rebooted and tried setting host
to each individual node, he would get the following error
message: %system-f-short.
SOLUTION: Permanent database was missing."
- "Getting aborted service request after reboot on the
server node.
SOLUTION: NCP showed no entry for satellite in database."
14 Digital Equipment Corporation - Company Confidential
14 Digital Equipment Corporation - Company Confidential
14 Digital Equipment Corporation - Company Confidential
- "Customer has a CI node and is having a problem making it
a BOOT server.
SOLUTION: needs to set ALLOCLASS = 1 to match HSC's ...
set NICS_LOAD_PEA0 = 1 to load the NI cluster driver."
Digital Equipment Corporation - Company Confidential 15
Digital Equipment Corporation - Company Confidential 15
Digital Equipment Corporation - Company Confidential 15
7 COMMUNICATIONS QUESTIONS
7 COMMUNICATIONS QUESTIONS
7 COMMUNICATIONS QUESTIONS
1.
1.
1.
What changes would you like to see implemented to simplify
communications management within a VAXcluster?
2.
2.
2.
When other people use you as an information source, what
questions are most often asked regarding VAXclusters communi-
cations?
16 Digital Equipment Corporation - Company Confidential
16 Digital Equipment Corporation - Company Confidential
16 Digital Equipment Corporation - Company Confidential
3.
3.
3.
When other people use you as an information source, what
information/advice do you most often give regarding communi-
cations issues?
Digital Equipment Corporation - Company Confidential 17
Digital Equipment Corporation - Company Confidential 17
Digital Equipment Corporation - Company Confidential 17
8 STORAGE
8 STORAGE
8 STORAGE
Peripheral devices can be configured into a VAXcluster to allow
local, or shared access. Both HSC and DSSI topologies make
unique demands on a Cluster Manager.
The following call samples highlight the types of problems
grouped in the "STORAGE" category.
NOTE
NOTE
NOTE
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
Statements in quotes ("Customer ran CLUSTER_CONFIG.COM ...
" ) are random calls and solutions taken from the CSC/CO
" ) are random calls and solutions taken from the CSC/CO
" ) are random calls and solutions taken from the CSC/CO
call database
call database
call database
8.1 Customer Questions Include...
8.1 Customer Questions Include...
8.1 Customer Questions Include...
- How do you define and differentiate Dual-Ported RA series
disks, and Dual-Ported MASSBUS and DSSI disks?
- "Customer wanted to know if he should answer YES to en-
abling a new satellite as a disk server.
SOLUTION: ... if other members need to reference the
disks, the answer is YES..."
- "4000-200 DSSI drives are not seen by any other nodes.
SOLUTION: The ALLOCLASS of the DSSI drives is set to 0 ...
needs to be set to 1"
- "Has 2 3800 and wanted to know about dual hosting.
SOLUTION: ... If the VAXcluster software is turned on and
the MSCP serving is enabled, then either of the 3800's
will be able to MSCP serve the disk ..."
18 Digital Equipment Corporation - Company Confidential
18 Digital Equipment Corporation - Company Confidential
18 Digital Equipment Corporation - Company Confidential
- ... two nodes, an 8250 and a 6510 ... with disks dual
ported ... several disks served to the 8550 from the 6510.
Why?
SOLUTION: The only way to get the path back to a preferred
path would be to write a program to select the path ..."
- "Customer is trying to add a satellite but CLUSTER_
CONFIG.COM says the system isn't a BOOT server.
SOLUTION: ... system disk is an Emulex disk an the HOST_
TYPE comes up as MD. Customer edited CLUSTER_CONFIG.COM
to recognize MD as HOST_TYPE and now CLUSTER_CONFIG.COM
continues."
Digital Equipment Corporation - Company Confidential 19
Digital Equipment Corporation - Company Confidential 19
Digital Equipment Corporation - Company Confidential 19
9 STORAGE QUESTIONS
9 STORAGE QUESTIONS
9 STORAGE QUESTIONS
1.
1.
1.
What changes would you like to see implemented to simplify
storage management within a VAXcluster?
2.
2.
2.
When other people use you as an information source, what
questions are most often asked regarding VAXclusters storage?
20 Digital Equipment Corporation - Company Confidential
20 Digital Equipment Corporation - Company Confidential
20 Digital Equipment Corporation - Company Confidential
3.
3.
3.
When other people use you as an information source, what
information/advice do you most often give regarding storage
issues?
Digital Equipment Corporation - Company Confidential 21
Digital Equipment Corporation - Company Confidential 21
Digital Equipment Corporation - Company Confidential 21
|
| 136.25 | Can't get there from here.... | MELKOR::SWIERKOWSKIS | | Wed Apr 29 1992 14:01 | 5 |
| Paul,
They need to fix protection/DEFAULT access to the files....
Susan
|
| 136.26 | File Access and Various. | HARDY::MOSTEIKA | Paul, ZKO1-1/D42, 381 (881)-1075 | Fri May 01 1992 12:28 | 28 |
| O.K. I sent mail to Harold Owens regarding access to the VAXcluste Questionnaire.
On the other:
I hear what your saying regarding Quorum. I think it's pretty important too.
I meant in reference to what was covered in SYSNET. The VAXc course
has the same coverage as SYSNET minus the troulbleshooting piece that comes at the
end. You are obviously adding from your own experience. This is great
for the students. However if your trying to squeeze that into the SYSNET
lecture then there isn't time. There has to be a tradeoff - other stuff
will have to be dropped or de-emphasized.
As far as quorum in the new VAXc course - the two day will have the
SYSNET theory stuff, the three day will have the troubleshooting. I should say
that that is my understanding, (I'm not working on that section). The
troubleshhoting piece that's in the old VAXc course is left pretty wide open.
You'll have to elaborate some time about depositing the SIRR from the console.
I said WHEN you get a chance. I've taught thirteen week quarters, and it sure
wasn't any fun. Ten or eleven was typical. What ever happened to those 66%
platform promises?
I'd like to see if we can incorporate your lab, if that's o.k.. I'll check with
the others.
Regards,
Paul
|
| 136.27 | | AUBREY::DONHAM | Progress Through Tradition | Fri May 01 1992 15:32 | 6 |
|
Wasn't 'The Others' a 1950's sci-fi thriller?
Perry (one of the Others...)
:^)
|
| 136.28 | VAXcluster Problem Report/Questionnaire Location (NEW) | HARDY::MOSTEIKA | Paul, ZKO1-1/D42, 381 (881)-1075 | Fri May 01 1992 18:17 | 19 |
| From: SPEZKO::OWENS "Harold A. Owens ZK01-1/F22 Dtn: 381-1687 01-May-1992 151
2"
To: SUPER::MOSTEIKA
CC:
Subj: RE: File Access
Hi Paul,
The new location for the questionaire is:
GETJPI::SYS$PUBLIC:CLUSTERS_QUESTIONAIRE.PS
GETJPT::SYS$PUBLIC:CLUSTERS_QUESTIONAIRE.TXT
The system management team on CSSE32 virtually "killed" DECnet access to
SYS$PUBLIC.
-->Harold
|
| 136.29 | Believe it or not.... | MELKOR::SWIERKOWSKIS | | Fri May 01 1992 19:02 | 46 |
| > I hear what your saying regarding Quorum. I think it's pretty important too.
> I meant in reference to what was covered in SYSNET. The VAXc course
> has the same coverage as SYSNET minus the troulbleshooting piece that comes at the
> end. You are obviously adding from your own experience. This is great
> for the students. However if your trying to squeeze that into the SYSNET
> lecture then there isn't time. There has to be a tradeoff - other stuff
> will have to be dropped or de-emphasized.
I am not just trying to squeeze it in; I AM squeezing it in. It's amazing
how fast you can talk when you have to.... In addition to talking fast, I
certainly don't spend much time on forms for print queues, monitoring or
performance (and I don't do module 1). So, day one is DECnet (all day) and
day two (plus an hour on day three) is VAXcluster. I get the hardware and
software overview in AND my hour on Quorum. They do the cookbook lab and we
finish up with the fine tuning the next day. Are we all exhausted? You bet!
> As far as quorum in the new VAXc course - the two day will have the
> SYSNET theory stuff, the three day will have the troubleshooting.
From what I've seen of the first chapter draft, the two day VAXc cluster class
has a heck of a lot more theory/overview information than the SYSNET courses.
Is SYSNET III going to be overhauled to match?
> You'll have to elaborate some time about depositing the SIRR from the console.
I don't do much beyond the example in VAXc (p. 6-39) and SYSNET III (p. 4-42).
The difference is that I talk about Quorum all at once (booting and shutting
down). Over the years I've noticed that it cuts down on the amount of time
needed to justify Quorum if I give them the options up front (inevitably there
will be one student who thinks Quorum is stupid and can't see why we have to
have it; this way, even though they may still not like it, they can work
with it).
> I said WHEN you get a chance. I've taught thirteen week quarters, and it sure
> wasn't any fun. Ten or eleven was typical. What ever happened to those 66%
> platform promises?
HAHAHAHAHAHAHAHAHAHA!!!!!
Whew! Thanks, I needed that!?!?!
> I'd like to see if we can incorporate your lab, if that's o.k.. I'll check with
> the others.
Fine with me; saves me some time at the photocopier....
Susan
|
| 136.30 | What happens to cluster internals | JGODCL::BRINKS | | Fri May 08 1992 06:09 | 8 |
| Hello to you,
When building new currula for repair technicians we were looking
for the cluster internals course. Is this course still available,
will it be part of the sysnet string ??? What happens in the meantime??
Peter Brinks
Training Nijmegen Netherlands.
|
| 136.31 | Nothing happens, from our standpoint | SUPER::MATTHEWS | | Fri May 08 1992 13:51 | 17 |
| Hi Peter,
We don't maintain the cluster internals training here in VMS Course
Development, and it is not affected by the work we are doing.
The "sysnet string" is funded by Customer Training. They do not fund
any cluster internals course, and cluster internals doesn't appear
in the Sys/Net string for customers.
As far as I know, all the cluster internals training has been funded by
Digital Services Training. That organization is looking at adopting the
Sys/Net courses, but I don't know what progress they've made, so I
don't know whether the Sys/Net and cluster internals courses will ever
appear in the same curriculum. You might want to get in touch with
George Stanley, who is the VMS curriculum manager for Services training.
Val
|
| 136.32 | | GOONS::BAKER | What does "ignorant" mean? | Sun May 10 1992 11:09 | 23 |
| Re. Cluster Internals courses
We in the UK (also serving Europe) have recently split the original Cluster
Internals/Level 2 course (EY1735) into four.
EY-J135 - Systems Communications Architecture L/L 5 days.
EY-J136 - VAXcluster Software Internals L/L 4 days.
(Connection Manager, CSP, CWPS, overview troubleshooting)
EY-J137 - VAXcluster Disk I/O L/L 4 days.
EY-J138 - Distributed Lock Manager (Internals) Seminar 2 days.
These have been scheduled to run during Q1-93 as employee-only courses.
These are purely software courses. Hardware aspects are covered in
brief overview (if at all). The reason for splitting is to allow
material to be adequately covered (all of this in 10 days was too
much!) and to permit consolidation between topics.
J135 is a prerequisite for J136 and J137 - the latter two can be
taken in any order.
Stephen
|
| 136.33 | Capitol Ideas | TEACH::LYNN | | Sun May 31 1992 19:28 | 13 |
|
Please do not destroy the cluster course until you completely fix the
SYSNET series. How can you write a new 2 day/3 day cluster course
until you are sure what is being covered in SYSNET I, II, and III.
If you feel that the information in the SYSNET series is equal to
what is in the current Cluster class, Please look again. SYSNETII
has three (not so great) pages for the SHOW CLUSTER Utility. The
Cluster Course does a much better job in six pages. Even more
could be added, with labs to reinforce learning.
Lynn White
Washington, D.C.
|