Re: RSX-11M-Plus Pool questions
- From: "Bob Gezelter" <gezelter@xxxxxxxxx>
- Date: 27 Mar 2007 04:26:19 -0700
On Mar 27, 5:14 am, "Duncan Macdonald"
<duncanmacdon...@xxxxxxxxxxxxxxxx> wrote:
If you are worried about the impact that the TCP/IP software might have on
your RSX-11M+ systems then the cheapest answer might
be to use a Linux box running both TCP/IP and DECnet to act as an access
point to your network.
(Seehttp://linux-decnet.sourceforge.net/for details of the DECnet support
available to Linux)
An alternative might be to buy a cheap Alpha VMS system (e.g. DS10L) from
someone like Island Computers (http://www.islandco.com/)
with DECnet and TCP/IP to act as the bridge but this will cost a few
thousand dollars with the licenses.
"Jeff Cameron" <rokt...@xxxxxxxxx> wrote in message
news:C22DAB95.28536%roktsci@xxxxxxxxxxxx
Since there is no comp.os.rsx group, and with all the old timer's, like
myself that have begun their VMS zealotness with PDP-11's and RSX, I am
posting this request for information on this comp.os.vms board. If you are
not interested in RSX-11M+, I would suggest you ignore this post.
First, a little background. After leaving MTI, I have now landed another
job
with a company that uses PDP-11s running RSX-11M+ to monitor and track the
manufacturing of product lots. I am actually performing backups using BRU
to/from RA-92 disks and 9-Track tapes. While I know I will be receiving
many
responses of condolences, both sincere and sarcastic, they are paying me
well. In fact, very well! I have given myself the title of
"Paleocybernetic
Analyst". The fact that these 4 PDP-11 systems are still running after 30
years, is a testament to the quality and reliability of DEC equipment. In
the past 10 years, the only hardware problems that we have had are a few
dead DZ-11 boards.
I'm sure many of you will respond with "Why doesn't your company migrate
to
a more modern platform?" Don't worry, we are! But this migration is
embedded
into a complete factory floor modernization, complete with modern
manufacturing process monitoring and control systems, who's "Go-Live" date
is slated for late 2009 or Early 2010. My job is to maintain the 4 PDP-11
systems through that time.
Because of the nature of the company's business, our operations are bound
and monitored by Federal Regulations, so the data collected in the
manufacturing of the product must be maintained and accessible for 15
years.
So besides maintaining the hardware for another 3 years, we will have to
maintain the data and be able to provide the ability to analyze the data
after the PDP-11s are finally retired.
These PDP-11's are on a segregated DECNet network (thick ethernet with AUI
connections along with 1 VAX 4000 and several DECServer Terminal servers).
This DECNet backbone runs throughout our 3/4 square mile campus in several
buildings. Users who use the in house developed Reporting programs, which
are written in Macro-11, connect to the PDP-11 systems and single VAX, via
Serial lines connected from their PC's COM ports to these DECServers.
My immediate project is to provide IP connectivity to this network, both
for
Telnet and FTP connectivity. Fortunately, Process Software does have
TCPWare
for the PDP-11/RSX. Not only will this allow us to remove the DECServers,
and localize the dedicated DECNet ethernet to a DELNI in the computer
room,
but it will give us connectivity to our large SAN storage on the corporate
LAN. Using FTP, I can then store the binary RSX data files (which are
currently offline on racks of 9track tapes) online and available to users
from their PCs.
Here is where I need your help. First, I need a network bridge; one that
does AUI to either 10-Base-T, 100-Base-T, or Fiber. The bridge only needs
to
bridge IP traffic and not DECNet. Any recommendations for such a bridge
wold
be helpful. Since the Dedicated DECNet network only does 10 Megabits, a
10-Base-T connection would be more than sufficient.
My next question is more involved, as it pertains to available pool space
in
RSX. I have read all the RSX documentation I can find relating to pool
space, and I am still confused about a few things. I need to accurately
quantify the impact of the TCPWare on available pool space on one of our
PDP-11/RSX systems. I know that there are two types of pool in RSX,
Primary
and Secondary. I can quantify the situation of Secondary pool using the
MCR
command >SET /SECPOL or the DCL command >SHOW SECONDARY along with the
command >SHOW PARTITIONS. I think I have a good grasp on Secondary pool.
The TCPWare documentation describes its pool usage very clearly, however
it
uses the term "POOL" and "INPOOL" instead of "Primary Pool" or "Secondary
Pool". The TCPWare "POOL" requirements are 700 to 1200 (no units
specified).
The TCPWare "INPOOL" usage is defined as: <usage>=(4*n)+24 where "n" is
the
number of TCP connections and the answer is in KBytes.
Question 1 : Is "INPOOL" the same as "Primary Pool", "Secondary Pool", or
possibly something else?
Next, the RSX documentation states that the "Primary Pool" values can be
determined from the MCR command >SET /POOL. This yields the following
results:
POOL=1200:01150.:01538.
According to the documentation, the results shown are:
POOL=<top>:<max>.:<total>
Where:
<top> = The first location for user partitions (in bytes)
<max> = The longest free block (in words)
<total> = The total pool space (in words)
Based on the values shown from the SET /POOL command, I have 2300 bytes
(2.3
Kbytes) of free space which is more than an order of magnitude smaller
than
the TCPWare requirements, which say that just one connection would require
28 Kbytes. Also, the RSX programmer here that originally programmed the
system says we ar at the maximum of pool space alowed. If so that would be
1538 from the "total" value above. If TCPWare could take up 1200 of the
1538, then it would be a big pool hog. Clearly something (if not my
reasoning) is not correct here.
What if INPOOL refers to secondary pool? Here the RSX MCR command SET
/SECPOL results in:
SECPOL=804.:1024.:78%
According to the documentation, the results shown are:
SECPOL=<secfr>:<secsiz>:<pctfr>
Where:
<secfr> = Number of Free blocks in secondary pool (64 bytes/block)
<secsiz> = Size of secondary pool in blocks (64 bytes/block)
<pctfr> = Percentage of free secondary pool space.
Based on these values, if INPOOL refers to secondary pool, then I might be
able to accommodate up to 6 IP connections.
So I'm hoping INPOOL refers to Secondary pool.
Question 2 : In regards to primary pool, do I consider the <max> value as
the total amount of free primary pool space, even though it is listed as
the
"longest free block"?
One thing to note. I cannot enable the PMT (Pool Monitor Task), because I
would have to write up documentation as to how this would impact the
production system and go through a lengthy approval cycle, since it is not
currently part of the system. I am already having enough problems
attempting
to write the "Change Requirements" just to test the TCPWare.
Question 3 : What is the basic difference between Primary and Secondary
pool? Based on the chapter 8 on Memory Management in the RSX System
Management guide, I surmise that Primary pool is for internal RSX tables
and
data structures, and secondary pool is for shared application memory and
installed tasks. Is this a correct assessment?
Final Question : Does the RSX system Generation utility alow you to
increase
Primary Pool and/or Secondary Pool? If so what are the limits (maximum
pool
size) for each? The doumentation always shows the MCR command >SET
/POOL=1200 as setting the top value to 120000 (octal). Is this the
all-time
maximum alowed in RSX?
Thank you for your time in reading my long winded question, and any
insight
you might be able to provide. I have a strong background in VMS, so you
may
use comparisons and contrasts, if you wish.
Thank you
Jeff Cameron =
Jeff,
It has been a while, but I have done a significant amount of RSX-11
work (S/M/M-PLUS) in the past (down to the Executive and driver
level). And I am writing this from memory, without dragging out my
Executive Listing (still on the shelf in my library) or my
documentation set (co-located near the executive listings).
The restrictions on the two pools are based on the memory mapping
hardware of the PDP-11 processors. The mapping of primary pool must be
within the first five mapping regions (KISAR5-6 are reserved for
loadable executive components) and the lowest part of the memory map
is reserved for the Executive itself. Secondary pool is mapped
differently. As I am sitting here, I would have to review the listings
for the guidelines.
Secondary pool can get large (at least by PDP-11 standards). By VAX,
ALPHA, and Integrity standards, it is small (the limits of the PDP-11
memory space are, IMHO, the reason for the jokes about "Non-Paged
Ocean" occasionally heard at OpenVMS sessions).
Should you be able to support some TCP connections reasonably without
a problem, yes. Hundreds or thousands, probably not. Also, I would
have to check things, but I suspect that secondary Pool can be
expanded without serious problem. Secondary pool can be re-done
without a SYSGEN.
If I can be of further assistance, please let me know.
- Bob Gezelter, http://www.rlgsc.com
.
- Follow-Ups:
- Re: RSX-11M-Plus Pool questions
- From: B. Z. Lederman
- Re: RSX-11M-Plus Pool questions
- References:
- RSX-11M-Plus Pool questions
- From: Jeff Cameron
- Re: RSX-11M-Plus Pool questions
- From: Duncan Macdonald
- RSX-11M-Plus Pool questions
- Prev by Date: Re: Time for C.O.V.subgroups ?
- Next by Date: FC SAN UDID
- Previous by thread: Re: RSX-11M-Plus Pool questions
- Next by thread: Re: RSX-11M-Plus Pool questions
- Index(es):
Relevant Pages
|
