[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference irnbru::ims_api

Title:

Moderator:	IRNBRU::MACKENZIE

Created:	Wed Jul 10 1991
Last Modified:	Tue May 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	289
Total number of notes:	1527

289.0. "Problems with large request elements" by OZROCK::PERKINS (Some contractors, their sex-life destroyed and overwhelmed by project pressures, turn to snorting quack.) Thu Apr 24 1997 08:03

Hi Guys,

any ideas???  
btw what is that maximum element value size?
is it documented anywhere?
is there a constant in an include file??

Thanks Kevin

From:	OILS::"[email protected]" "24-Apr-1997 1034" 24-APR-1997 11:24:02.88
To:	[email protected]
CC:	
Subj:	Problems with large request elements - Configuration of IMS/DMQ?

We are sending IMS requests containing multiple elements with each
element approaching the DMQ 3.2 limit of 32000 bytes. Each time we 
do this IMS reports and error on the VAX:

* SL-I-TARGETFULL: Error sending to group_num:queue_num, trying 
again in 2 seconds *

This message is repeated until IMS gives up trying.

How do we configure IMS/DMQ to:

1) handle requests with multiple large elements; and
2) handle numbers of users (>500) and server applications (>10) that 
are likely to be sending several medium size (6000 - 15000 byte) 
messages per minute each?

I have tried reconfiguring the DMQ  buffer pool configuration table 
where I increased the number of 32000 byte buffers from 1 to 10. I 
have not yet increased the number of buffers for the other pools 
(small & medium) and neither have I changed the warning level or 
reserve count values for any of the buffer pool sizes. We have also 
increased the quotas for GBLPAGES, GBLSECTIONS & GBLPAGFIL by 50%, 
run SYSGEN and rebooted the VAX. None of these changes seems to have 
had any effect.

Am I looking in the wrong area? What other parameters should we 
change? Do the same changes apply to IMS/DMQ running on 
Alpha/OpenVMS?

Any/all suggestions would be gratefully received.

Thanks,

Neil
//---------------------------------------------------------
// Neil Muller
// Asst Mgr. Application Development,
// Attorney Generals 'D' Branch
// email: [email protected]
// voice: +61 6 234 1362
// fax:   +61 6 234 1480

T.R Title User Personal
Name Date Lines

289.1 MSG+ max element size NNTPD::"[email protected]" Dave MacKenzie Sat Apr 26 1997 12:10 65

T.R	Title	User	Personal Name	Date	Lines
289.1	MSG+ max element size	NNTPD::"[email protected]"	Dave MacKenzie	`Sat Apr 26 1997 12:10`	65
	Hi Kevin, >>what is that maximum element value size? 31500 >>is it documented anywhere? Should be in the release notes. >>is there a constant in an include file?? No >>I have tried reconfiguring the DMQ buffer pool configuration table >>where I increased the number of 32000 byte buffers from 1 to 10. I >>have not yet increased the number of buffers for the other pools >>(small & medium) and neither have I changed the warning level or It's more complicated than that. The basic problem is the queue does not have enough space left on it for the message. He needs to up the queue quota to be able to hold all messages waiting to be read. This is defined by either bytes or number of messages. Since it's VMS he also need to get the buffer pool config correct. Each message pending on a queue will take up a buffer. He must have enough buffers of each size to cope with the incoming messages, and the buffers should be sized correctly. By running the LLS Verify utility peridocally he can see the allocation at a given moment. This can be run as a seperate exe from dmq$exe, perhaps by a batch program. If he's connecting in a lot of PCs then he should also check his DMQ XGroup buffer pools to either the PC groups or Client Library Server groups.. There must be enough space in KB for any messages in transfer between the groups. This is less likely to be a problem. >>1) handle requests with multiple large elements; and Sizing details needed here. If he means a multi-chunk message then he'll need more large buffers. (each chunk will be around 30K) >>2) handle numbers of users (>500) and server applications (>10) that >>are likely to be sending several medium size (6000 - 15000 byte) >>messages per minute each? LLS Verify periodically to get to grips with this. >>reserve count values for any of the buffer pool sizes. We have also >>increased the quotas for GBLPAGES, GBLSECTIONS & GBLPAGFIL by 50%, >>run SYSGEN and rebooted the VAX. None of these changes seems to have >>had any effect. They won't unless he get's his DMQ config correct. Ask him for his DMQ$INIT if he's still having problems. >> Do the same changes apply to IMS/DMQ running on >>Alpha/OpenVMS? Yes. Also tell him to go on a training course. This is standard config stuff. Regards, Dave M. [Posted by WWW Notes gateway]

Hi Kevin,

>>what is that maximum element value size?
31500

>>is it documented anywhere?
Should be in the release notes.

>>is there a constant in an include file??
No

>>I have tried reconfiguring the DMQ  buffer pool configuration table 
>>where I increased the number of 32000 byte buffers from 1 to 10. I 
>>have not yet increased the number of buffers for the other pools 
>>(small & medium) and neither have I changed the warning level or 

It's more complicated than that. The basic problem is the queue does
not have enough space left on it for the message. He needs to up the 
queue quota to be able to hold all messages waiting to be read. This is 
defined by either bytes or number of messages.

Since it's VMS he also need to get the buffer pool config correct. Each
message pending on a queue will take up a buffer. He must have enough 
buffers of each size to cope with the incoming messages, and the buffers
should be sized correctly. By running the LLS Verify utility peridocally
he can see the allocation at a given moment. This can be run as a 
seperate exe from dmq$exe, perhaps by a batch program.

If he's connecting in a lot of PCs then he should also check his DMQ
XGroup buffer pools to either the PC groups or Client Library Server groups..
There must be enough space in KB for any messages in transfer between the
groups. This is less likely to be a problem.

>>1) handle requests with multiple large elements; and

Sizing details needed here. If he means a multi-chunk message then 
he'll need more large buffers. (each chunk will be around 30K)

>>2) handle numbers of users (>500) and server applications (>10) that 
>>are likely to be sending several medium size (6000 - 15000 byte) 
>>messages per minute each?

LLS Verify periodically to get to grips with this.


>>reserve count values for any of the buffer pool sizes. We have also 
>>increased the quotas for GBLPAGES, GBLSECTIONS & GBLPAGFIL by 50%, 
>>run SYSGEN and rebooted the VAX. None of these changes seems to have 
>>had any effect.

They won't unless he get's his DMQ config correct. Ask him for his 
DMQ$INIT if he's still having problems.

>> Do the same changes apply to IMS/DMQ running on 
>>Alpha/OpenVMS?

Yes.

Also tell him to go on a training course. This is standard 
config stuff.

Regards,
Dave M.

[Posted by WWW Notes gateway]