tcpip proxies

[Gregg Parmentier]

VMS 7.2-2
tcpip 5.1

"$ tcpip sho config comm" shows 2500 max proxies
"$ tcpip sho comm" shows 429 max proxies

What's causing the running system to not use the configured value? Is there a sysgen parameter or an account limit that I need to increase?

Why doesn't "$ tcpip sho comm" show any values for current and peak proxy use?

[John Gillings]

Gregg,

V7.2-2 is an orphan version. There could be all kinds of long-since-fixed bugs that everyone has forgotten in the years since V7.3 superceeded it.

Similarly TCPIP V5.1 is way off the radar.

(take it as read the usual stuff about the myth of application code which runs on V7.2-2 but is unable to run under V7.3-2)

To your specific questions:

>What's causing the running system to
>not use the configured value?

Has TCPIP been restarted since the new value was set?
You may want to try reloading the proxy data base with:

$ TCPIP SET TCP/SIGNAL

>Is there a sysgen parameter or an
>account limit that I need to increase?

I don't think so. TCPIP limits belong to TCPIP only.

>Why doesn't "$ tcpip sho comm" show any
>values for current and peak proxy use?

Good question. I see the same on V8.3-1H1 and TCPIP V5.6. I suspect the display code is using a generic template shared with other attributes that track current and peak values, but proxies don't. It's not as if they typically fluctuate.

[Gregg Parmentier]

It's been rebooted since I made the change. Before the reboot the maximum was listed as 533 without a maximum set in the configuration.

I see that 7.3-2 with tcpip 5.4 seems to have correct values. But I can't find patches for 5.4 to fix the pseudo-interface problems.

Will tcpip 5.6 work correctly with 7.3-2?

[STEPHEN CONWAY_1]

> Will tcpip 5.6 work correctly with 7.3-2?

No.

Due to IA64 port (prived structure changes).
Use the latest V54 eco (V5.4-ECO7).

I am not sure what you mean by
"pseudo-interface problems". Looking back in
our problem history the last such problem
was fixed in V5.4-ECO3

Steve

[Hoff]

There are no significant difference in the kernel between OpenVMS Alpha V7.3-2 and OpenVMS Alpha V8.3 as far as typical applications will find; there are additions and such.

In practice, the OpenVMS Alpha kernel probably saw more significant perturbation between V7.3-1 and V7.3-2 with the TQE-related changes.

Yes, you might need updated device drivers or a device driver re-link pass, but how many sites are hauling around local device drivers anymore?

In other words, I'd not stop at V7.3-2 here; I'd go all the way to V8.3. (Just between you and me here, you're planning on staying on the OpenVMS release you upgrade to here for a while, right? Might as well get as close to current as you can manage when the schedule and the usual administrative shackles permit movement.)

Stephen: I vaguely recall seeing a version matrix around; what was supported where, and such. Yes; here's something similar:

http://h71000.www7.hp.com/network/tcpip_matrix.html

That doesn't list the ECOs, but it's close.

[Gregg Parmentier]

> I am not sure what you mean by
> "pseudo-interface problems". Looking
> back in our problem history the last
> such problem was fixed in V5.4-ECO3

Where do I get ECOs for 5.4? I've got the copy of 5.4 that came with the 7.3-2 distribution disks. The ECOs I can find on line (in ftp://ftp.itrc.hp.com/openvms_patches/layered_products/alpha/) are only for the Japanese version.

[Duncan Morris · Accepted solution]

Cunningly hidden in the main 7.3-2 folder

ftp://ftp.itrc.hp.com/openvms_patches/alpha/V7.3-2/


Duncan

[marsh_1]

gregg,

try here :-

http://www13.itrc.hp.com/service/patch/search.do?searchOsid=openvms:alpha:7.3-2&pageOsid=openvms&searchType=patch.text.searchTypeBrowse&pageNumberStr=1&pageSizeStr=25&BC=main|

good luck

[Wim Van den Wyngaert]

I have 7.3 with tcp 5.3 eco 2.

On an AS500, I tried to put the number of proxies on 100.000 but after restart it was on 16960.

Did the same on a AS100. Got 34464.

So there must be something that limits the number and it's not the version of tcpip.
I tried to play with pgfl, wsext, bytlm. Nothing. show mem/fu/all reveals nothing. scanning the tcp exe files idem. sysgen param values idem.

???

Wim

[Wim Van den Wyngaert]

Tried again on the AS500. This time it got 34464. And I didn't change anything else (and I verified the screen contents : my previous tests did return the value I said).

I rebooted instead of restarting tcp and got 34464.

AI ?

Wim

[Wim Van den Wyngaert]

Tried it with all kind of values and until 65.000 they were accepted.
But with 65.537 I only got 1.
And at 80.000 I got 14464.

But your version is different. Just try all kind of values until it works.

Wim

[Volker Halle]

Wim,

this looks like a display problem or an integeger (word) overflow problem.

100000. = 0x186A0
34464. = 0x86A0

Volker.