home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
OS/2 Shareware BBS: 8 Other
/
08-Other.zip
/
cnfgls30.zip
/
NEWS.TXT
< prev
next >
Wrap
Text File
|
1994-02-23
|
22KB
|
445 lines
CONFIG LS 3.0 -- News of changes and additions.
Version 1.00 Changes from CNFGLS13
There are a lot of these.
1. Additional questions on page 1 are included so some tuning
can be done depending on whether you have the Advanced or the Entry
server.
#3 asks which you are using -- easy enough -- but watch out for #3a!
#3a asks if FAT files are shared (if you are using the Advanced server)
or it asks if HPFS files are shared if your are installing the
Entry server. Caching (CONFIG.SYS) statements change depending on
your answers, so be carefull of the changing question!
#3b asks if you are configuring a Domain Controller or an Additional
Server. A DC takes a bit more RAM with 3.0 than the AS.
#11 While we are on memory requirements, the new OS/2 RIPL feature
takes more server RAM than the DOS RIPL we have had in the past.
So, question #11 has been Cloned! 11a wants to know if you have
any DOS RIPL requesters, and 11b wants to know about any OS/2
RIPL requesters. That is all the new stuff on page 1.
2. The second screen of questions is new and asks for information
about your LAN adapter(s). You may have none (and use the loopback
driver) or you may have up to 4. If you have adapters, they may be
Token-Ring or EtherNet. If they are Token-Ring, they may be 4Mbps
16/4Mbps or Busmaster adapters. If they are EtherNet, they may be
from any of the manufacturers whose cards are supported. Answer
the questions and review the generated parameter recommendations.
These appear in Column D - to the right of the questions (the same
place all those pesky error messages show up.)
3. Screen 4 has more error messages than before, but you still need
to ensure you have a string of OK's displayed here before
going on to the results. OK, I know the last entry will say
"None" if there are no warnings. The OK's are the important things.
4. There are several new error checks and error messages. These
should not make use of the tool harder, though.
5. Note that some new fields are calculated. Primary/ALTERNATE adapter,
adapter buffer size and some NTS/2 parameters to name some.
6. The Transmit Buffer Size recommendation changes depending on your
adapter type. Don't worry, it is not an error.
7. I tried to include the new memory requirements, but these are taken
from the documentation, and are not something I've tried out myself,
so I don't swear by them. You shouldn't either. Use of this tool
to find a minimum system is dangerous! It will give minimum
performance, and it may not run at all.
8. You might notice that there are fewer assumptions made now.
One that is important is that if you have a Busmaster TRN adapter,
I assume it is the 32 bit adapter. If you have the 24 bit version,
you might have a problem. The 24 bit adapter will not support
memory over 16MB, but the spreadsheet might try to allocate cache
there. Your generated configuration won't work right.
9. As indicated, the Work Area and the Error messages areas have grown.
10. There may be some blank lines at the bottom of the INI file that
could make you think you have seen all the generated stuff. Don't
be fooled! Page down through them until you see the END OF DATA
message. The blank lines would be used for configuring adapter
cards (NIC's) you do not have.
11. At he very end (bottom) of the spreadsheet are the cache
recommendations. THESE REQUIRE MODIFICATIONS BY YOU.
This may be just changing them to add the correct drive letters,
but if you have other work to do, there will be messages that appear
to tell you (123 users) what it is -- so read the messages!
It is possible to use CNFGLS13 on a 3.0 server, but use of CNFGLS30 will
probably give a better performing system. Using CNFGLS13 can also
waste some memory. Will run, though.
-------------------
Well, I intended to update this, but maybe not quite so soon.
If you know somebody who has the package but may not have subscribed
to it on MKTTOOLS, please let them know Version 1.01 is now current.
Two changes.
1. The IBMLAN.INI file's 'netx' line had 'lm10' specified. This
must be capitalized (LM10). Now, all alpha characters in the 'NETx'
line(s) will be caps.
2. The 'netbeui_nif' Sessions, Commands and Names in the PROTOCOL.INI
file must be at least one greater than the IBMLAN.INI NETx line
specifies. Now, they will be.
-------------------
Well, another week and another set of changes. I need to thank Ken
Whitfield and all the folks he got together with me so we could get
the input of the LAN Server and NTS development teams. A few days
in a conference room generated a lot of detailed changes -- to many
to itemize here. You'll see the differences, though, if you inspect
the results of a run at all. This should give better performance than
other, earlier, versions of CNFGLS30, so make sure folks know about it!
So, here is Version 1.02
4/9/93
--------------------
Made it two weeks without an update! Must be getting better at this.
These changes are all from user feedback on Prodigy and the OS2LAN CFORUM
so I thank you folks for the help! After all, we're all in this together
and I sure DO need the help!
1) I had an erro in calculating the 802.2 storage requirements that
might have given you a message about not enough work area that you
should not have gotten -- fixed.
2) 'Other NetBIOS Apps' resource requirements got added in to the LS
NETx line(s). Opps.
3) I was only allowing 5 802.2 users. Now, you can have 12.
4) The multiple adapter area was not fully cleaned up when you set up
for several adapters and then decided to see what would happen with
fewer. Now, I think everything you don't want to see disappears.
5) One user reported that the large number of maxsearches generated when
DLR's were supported gave problems with printing. Dunno, but I set
the generated value down, and sent the note on to the developers who
suggested the higher value. Maybe we'll see a fix to LS30 someday
and be able to increase it for those DOS apps that need higher values.
If your apps need the bigger number (it will be 700 now), just change
it in your INI file. No other parameters are affected.
4/23/93
-----------------------------------------------------
Well, looks like it is time for 1.04. Only a couple of things to change,
but there are over a thousand users of this tool that I know about, and I
certainly want to keep it as up-to-date as possible! 6 weeks with no
problems, but now I have 3 changes to make.
1. You could get a recommendation for additional RAM when you had
enough. A bug calculated a bad cache size which was detected as
being too small and (bingo) the recommendation popped out.
Adding the recommended RAM would make everything work out fine.
Looks like CNFGLS30 was trying to fix its own bug here, I dunno.
2. The 123 version omitted the RUN= statement in the CONFIG.SYS
recommendations to run the OS/2 Standard Edition (SE) cache.
It has now been added.
3. The Excel version specified this same cache should be loaded off
the C drive. Meant to say 'c' so you'd know to specify the drive
where OS/2 was installed. Finger check! Changed it.
Had a request to make FAT DISKCACHE optional, but that generates
several problems. Therefore, the recommendation is still there, but
you can just ignore it and add its value to the HPFS cache. This only
applies to Entry server configurations using OS/2 SE HPFS on all drives.
FAT (and DISKCACHE) is optional for Advanced server configurations.
6/8/93
-----------------------------------------------------
OK, on to 1.05. Discovered there were some notes posted where I cannot
see them that reported errors I knew nothing about. Thanks to you who
pointed them out to me! Fixed them and neatened up some other stuff.
1. Cell B75 was calculating the total number of NetBIOS Commands required
incorrectly. Have modified it to use the correct calculation, H14.
2. D28 and D33 calculated the MinRcvBuffs incorrectly on a 16Mbps Token-
Ring if you used a LAN Streamer adapter.
3. M11 did not allocate enough 'maxconnections' for users of the
Advanced server who also had FAT drives shared -- it allocated only
one connection, now it allocates 4 per user.
4. A98 (C98 in 1-2-3 version), etc., could ask you to configure more than
255 commands in a NETx statement and give you an error message which
might be ignored. Now, the maximum number of commands is taken
from H23 and will not exceed 255.
5. B75 was the error message that could have been ignored -- it now
you to H14 to see the total number of commands required. Adding
the recommended adapter card(s) can increase the performance of the
server, so you might want to look into doing that!
6. The PROTOCOL.INI area (where you define the number and type of
adapters) has been reworked, but you may notice no chcnges.
Mainly, there are new/changed error messages in Col D.
1-2-3 users should make sure you check Columns D and E for some
Recommendations that you might be missing. They may not display on
your screen. Just ignore the 'mystery numbers' in E78-83, it is just
a work area I use for debugging. A78-83 identifies them, if you care.
6/22/93 (First full day of Summer)
-----------------------------------------------------
11/01/93
Version 1.06
The new version of CNFGLS30 has many changes, but users will not be
aware of most of them. Comments I have received have generated almost
all changes, so please keep 'em coming!
1. Check out Question 3a carefully. In the past, it has asked if you
wanted to support the 'other' file system -- HPFS for Entry servers,
or FAT for Advanced servers. Now, it asks specifically whether you
want to share HPFS, FAT or Both. Your response will affect the new
(reworked) cache calculations. Those of you who install the Advanced
server and want to use the standard HPFS386 should note -- you will
be running the Entry server, so just specify it in Question 3. Also
might want to buy it and save some $$$.
2. Question 13a is new. It asks how much memory you want CNFGLS30 to
leave free on your server. Right now, it leaves 1 MB of RAM free, and
that is the default (and minimum acceptable value.) You will want to
increase this number if you have any apps running on the server (NOTES
or whatever). In the past, you had to subtract this reserved memory
from cache values to free it up since CNFGLS30 assumed the server was
the only product using the system. Not making the required adjustment
causes swapping and that is Not good for server performance.
By the way, if you are thinking about another program using the server
and wondering how it will affect server performance, remember that the
server is a heavy user of 1-the LAN Adapter(s), 2-disk and 3-RAM (but
that is a number you can now control.) If you add another app on the
server that really needs a lot of these resources, look out! But, if
your app needs CPU (which the server doesn't totally use - especially
if you have the LANStreamer TRN adapter), you may not impact server
performance at all. Functions like 3270 Gateway and/or DDCS are CPU
users, mainly, so they might be suitable for your server. The
decision is yours, of course, but a little added information doesn't
hurt, 'eh? OK, enough of that, let's move on.
3. Question 17 is also new. It asks for the number of 4033's you want
to support on the server. The 4033 attaches directly to the LAN and
supports one or two printers which appear to the server to be locally
attached. With 4033's, the server can support up to 24 printers. It
has been a bit difficult to get good, reliable information on its
requirements, so I can see why so many of you asked for this addition.
I certainly hope this does what you wanted it to do. The generated
parameters may well work for other LAN attached printers, too. At
least, they seem to handle one H-P user OK.
4. In Cell A29 (and the other three which repeat the message) you may
notice that you are no longer asked about the Busmaster adapter.
Now, you are asked if you have the LANStreamer. Respond with an 's'
in B29 if you do. 'B' is no longer accepted as a valid response.
Actually, CNFGLS30 always has configured a LANStreamer adapter when
the 'b' response was valid. So, only the confusing question and
response have changed. The PROTOCOL.INI stuff generated in Col D
stays the same.
5. Many of you have asked what the requirements were for support of
LOTUS NOTES. Well, calling LOTUS didn't help a lot since they didn't
seem too sure, either! So, I've gathered information that several of
you have sent me (Thanks!) and you will see another new question in
Cell B52 - 'LOTUS NOTES on server?'. A 'y' response will then cause
another question - 'How many concurrent users?'. Answer both, and
CNFGLS30 uses our combined experience/wisdom to allocate the resources
that should be required to support that product. If you have a
problem with this, as always just let me know so a change can be
made. It is probably right now, but if it isn't, it'll cahnge in 1.07
Oh! I almost forgot -- if you have NOTES, Col D will remind you to
go back to Cell B14 to allocate RAM for it there. I cannot change
your responses, so you gotta do it.
6. The LAN Network Manager requirements have been changed -- used to be
the old product, but now the real thing is in there.
7. NetBIOS Commands required could get large as the number of users
increased. It is now limited to a maximum number -- the number of
server buffers available for the data the commands work on.
The error messages screen has one change. The NetBIOS Commands (ncbs)
number which is generated by CNFGLS30 is high -- for better
performance. But, you can easily get a warning that the adapter
card cannot support the required number of commands. Well, it is
now a kinder, gentler warning -- getting it does not mean your
configuration will not work. Getting another adapter could give you
better performance, though.
The Commands value has caused some users problems because of this
error message, so I believe an explaination is in order. In past
versions of this tool, the number of commands was limited to be no
greater than the number of buffers allocated (srvpipes, reqbufs, etc).
In the first versions of CNFGLS30, this was changed to be 1.6 commands
per user, and that could be a lot of commands. Thus the error message
you see (really a warning as we all understand now, right?).
Now, the Commands value defaults to 50 , but may be 1.6 times the
maxusers value or the number of buffers or 450 --whichever is
smaller. Should decrease the number of times the error message shows
up. In NO case will the final number generated be larger than the
number of commands your adapter(s) can support, tho.
8. The number of Commands in the NETx lines of the IBMLAN.INI file is
now 1 less than the number generated for the NetBIOS configuration.
So, you no longer need to add one like you did in 1.05. Not a problem
for all users, but some have seen (and reported) it - so the 'fix'.
9. Server heuristics digit 14 has changed -- it used to be set to a 1,
but now reverts to the default of 0. So, you'll have to use
backslash characters as required since this digit will no longer
allow use of forward slashes and convert them to backslashes.
Saves a few cycles and might help performance.
10. The HPFS386 IFS statement recommendation for the CRECL parameter was
too large, and it now has a maximum value of 32.
11. It was possible to generate a cache size recommendation that was an
odd number. Should always be even for best performance, and now it
will always be even.
12. You'll also notice a new RUN= statement for the HPFS cache. The
addition is to suggest you use a MAXAGE of 10 seconds instead of
the 5 second default. This can help the performance of busy servers.
But, since data can be in cache longer without being written to disk,
you may want to stick with the default. Once again, the choice is
yours. If you have bad power and no UPS, don't change what you've
got since loss of data will be more likely in the event of a failure.
13. Those of you who look beyond Col D will notice that Col H now gives
the PROTOCOL.INI file's names for the various NetBIOS parameters
and the 802.2. parameters. Where different, the real names are in
parentheses, (), after the old name -- or a contraction of it that
allowed everything to fit in the Col.
14. Again, note that the total NetBIOS requirements are divided equally
among the adapters you have, so you may see a high number in H13-15.
Don't let it worry you -- the final recommendations are in the
'{netbeui_nif}' cells below these -- and those numbers may be
different.
15. The 'Additional Server RAM Recommended' message will now appear if
there is less than 2MB of cache in the server.
16. Users of the Token Ring 16/4 adapter may have noted that the
PROTOCOL.INI recommendation for MaxTransmits was 50 -- and 49!
Of the two given, 49 was the correct one, and I have deleted the 50
that was causing some confusion.
17. Cells W34-42 have been added to support the new cache calculations
and the use of FAT or HPFS or both on a server. Calculations got too
complex for a single cell to hold.
18. You'll also see some new error messages in AE72 and 73.
19. Finally, the date of the last mod I make will be in Cell C2, and
My IBMMAIL ID (USIB22M6) is now in the heading. Should be easier for
y'all to get notes to me now. I'll respond on a time available basis
as usual -- this is a spare time (?) effort, after all.
-----------------------------------------------------
11/20/93
Version 1.07
1. When I added NOTES support, I had the choice of two different drivers
to support --- and I chose the wrong one. OK, 1.07 will now support
the NetBIOS driver. Thanks go to Josh in Cincinatti for his patience
with me -- and for his good work in the testing (and modification) of
the 1.06 version.
2. 4033 users liked the support for that device, but HP users didn't get
any help since the requirements of the two devices (as far as 802.2
resources are concerned) are different. Now, the question asks for
the number of LAN attached printers (4033 OR HP). The allocated
resources reflect the worst case between the two printer attachments
(the maximum 802.2 resources required for either) so that it makes no
difference what type of printers you have.
3. 1-2-3 users reported a locked cell. B55. Mia culpa. Unlocked now.
4. IBMLAN.INI maxopens value was incorrectly calculated for some Advanced
Server configurations with Both FAT and HPFS used.
5. The Additional Server RAM Recommneded Message appeared when there was,
in reality, plenty of RAM. Fixed.
6. Messages notifying you that apps (Notes and 4033 AES/HP print support)
must have RAM reserved for them (in Cell B20) now will not appear if
you have allocated RAM for them. Sorry, I don't know how much they
take -- and I don't know how to tell 4033 for HP print, anyway.
7. A new error message appears if you reservre more RAM than you have
available in the server for other applications in B20.
8. Advanced Server users now get more Commands and reqbufs. Two Commands
instead of 1.6 for Entry users, and 3 buffers instead of 2. Keeps
Entry server's RAM requirements low, and gives better performance for
Advanced server users.
9. CM/2 SNA Gateway support has been modified to include support for the
Distributed Feature (which allows users to use a copy of the product
on the server's disk, not theirs -- and saves 14-19MB of disk per
requester). Use of this feature requires a LOT of open files at the
server. You'll notice that if you have Entry server (or if you say
you support FAT files on an Advanced server) that the 'maxopens'
value goes WAY up. If you put the code on an HPFS disk on an Advanced
server that also has FAT disks, run the calcs twice (once specifying
no FAT disks to get the 'right' maxopens number, and once saying you
have FAT disks to get all the rest of the parameters calculated.
10. Since APPC is required by CM/2 Gateway support, but does not require
a link station in that case, 0 or blank will now be accepted in the
number of link stations field for APPC.
11. A couple more error messages have appeared to support the new
functions, so if you go exploring in Col AE, you might see them.
They will be found in AE74 and AE75.
12. LAN Streamer allocations reportedly caused error message.
Incorporated change suggested by user.
1-2-3 USERS. Be sure you watch Col "D" to see these messages!!
That is where they display as needed, but you may not see the column.
Review it for each of the first four screens.
-----------------------------------------------------
As always, your notes and comments are welcome -- suggestions, too.
I can be reached on PRODIGY (BCJS75A) Or TOMGORDY at AUSVM1 for PROFS
users. Internet TOMG@VNET.IBM.COM and IBMMAIL(USIB22M6).
TomG