TCP/IP Limitation?

[Jorge Cocomess]

Greetings,

Is there a limitation on the OpenVMS 7.3-2 with number of TCP/IP connections? If so, where and how could I check out this??
Background:
About 200 client connect through the application runnning on OpenVMS, but before they can connect to this application they must first connected through a Windows 2000 server acting as a Cache server.

Thanks,
J

[Arch_Muthiah · Accepted solution]

Hi J,

I thing that there are no limitations on a server's ability to handle millions of TCP connections if memory resources are available to service the connections. If memory is insufficient, the server will reject new connections.

Google says 5 connections can be queued per port for FTP connections.


Archunan

[Andy Bustamante]

There are limits on the number of BG devices(raised with TCPIP 5.4). Which TCPIP product are you running?

Assuming TCPIP 5.4, shipped with 7.3-2 there is a default 9999 BG allowed. You can increase this manually or by loading the tcpip scalable kernel. See
http://h71000.www7.hp.com/doc/732FINAL/TCP_RN/tcp_rnpro.html#scalable_kernel

You can also manually raise ovms_unit_maximum with the sysconfig utility. See the TCPIP Tuning and Troubleshooting manual at http://h71000.www7.hp.com/doc/732final/6631/6631pro.html

Is there a specific performance issue?

Andy

[Jorge Cocomess]

Andy - No performance issues at the moment. We're seeing drop connections from the client server (W2K) - We have about 200 connections through this client server and then pass-thru the application server (VMS). And yes, we currently using Pathworks Advanced server with license for 200 units.

Thanks,
J

[David B Sneddon]

Jorge,

Exactly what type of connections are dropping?

Dave

[Jorge Cocomess]

Dave - That would be TCP/IP connections that are dropping - Sometime 2-7 drops during the day.

Thanks,
J

[David B Sneddon]

Jorge,

Yes, I guessed it would be TCPIP... :-)
Are the sessions telnet, ftp, connections to
pathworks shares?

Dave

[Jorge Cocomess]

Dave - It's not FTP session. So, I guess it would be telnet. I will wait later today or when some users connect through the client server, I'll run netstat -n to see their connections and let you know more on what type of connections they are.

Let's say it's telnet, is there any tweaks we can do?

Thanks,
J

[Jorge Cocomess]

Dave - After I ran netstat -n I see al lot of NetBios connections. Any ideas?

[Andy Bustamante]

Is this a case of having to reboot the Windows 2000 system to restore Pathworks connections? If you're running Pathworks 32 ver 7.4 on the client there is an updated DECDNP114.SYS available from HP which corrects this. If you check the Pathworks event log, look for:

Added 32 packets to the datalink buffer pool, total=xxx

After xxx hits 512, the Pathworks connections will hang.

Andy

[Arch_Muthiah]

Jorge,

As I said earlier, the number of connection is depend on your VMS system available resources.

But you can limit the number of connections if you don't have enough system resources using CHANNELCNT SYSGEN parameter.

TCPIP> SHOW SERVICE TELNET /FULL shows the current setting/limit details, the default I thing it is 256. Changing this values requires enough system resource especially (nonpaged pool size)

If you want you try change using
TCPIP> SET SERVICE TELNET /LIMIT=n
make sure the value of "n" would not exceed the channelcnt SYSGEN param.

Because "too many sessions" error will come whenever we don't have enough non-paged mem.

I would suggest better contact Hp on this.


Archunan

[Arch_Muthiah]

J,

I just saw this from the ask the wizard, it says something about your window side client lic limit and WINdows sometime misinterpreting the client connections lic limit (pthwork client), try go thru this...

http://h71000.www7.hp.com/wizard/wiz_4432.html

Archunan

[Jorge Cocomess]

Here's the result after I ran "SHOW SERVICE TELNET /FULL"

Service: TELNET
State: Enabled
Port: 23 Protocol: TCP Address: 0.0.0.0
Inactivity: 1 User_name: not defined Process: not defined
Limit: 50 Active: 22 Peak: 35

File: not defined
Flags: Listen Rtty IPv6

Socket Opts: Keepalive Rcheck Scheck
Receive: 3000 Send: 3000

Log Opts: Actv Dactv Conn Error Logi Logo Mdfy Rjct
File: not defined

Security
Reject msg: not defined
Accept host: 0.0.0.0
Accept netw: 0.0.0.0


Please let me know your feedback.

Thanks,
J

[Robert Gezelter]

Jorge,

You noted that the telnet connections are limited to 50, and the maximum reached is 35, so that is unlikely to be the issue.

With that number of clients, you are not lilely to be hitting the 9999 units limitation on a daily basis.

I would, however, question one premis of this entire thread: That the OpenVMS system is the source of the problem.

If this is indeed happening several times a day, out of hundreds of connections, it is a rare enough event to be caused by other problems. To track this by science, rather than by guesswork, I would use a LAN monitor (e.g., Ethereal or Etherpeek) to capture and log the traffic that is having the problem. I would then locate the sessions that are being dropped and see exactly what DID HAPPEN not what MAY HAVE HAPPENED.

The logs will show the facts, not guesswork as to what could have happened. It is more grind, and more tedious than just adjusting parameters, but it is guaranteed to produce an accurate result that can be taken to management.

- Bob Gezelter, http://www.rlgsc.com

[Hans Blom_2]

Hi,
we had more or less the same problem, but with ftp. We found out a lot of server processes were in RWMBX and then connections timed out.
Raising the TQElm for account TCPIP$FTP solved the problem.

Regards
Hans