JTQUOTA

[Sk Noorul Hassan]

Hi all,
Thanks for your suggestions for JTQUOTA. my present JTquota for the user responsible for running almost 150 processes on VAX 4105/128MB is 1024. If I go ahead and double the value, what would be the impact on the system performances and other resources. Will the increased value will be taken from other pool or will have no impact on other resources.

Pls suggest ...

[Wim Van den Wyngaert]

It is taken from the paged pool.
So, 150 processes * 1024 extra memory is about 150K.

With show mem/pool check if the pool has this amount free space. If not, increase pagedyn.

Wim

[comarow]

JT quota comes out of pagedyn

show mem/full/pool

if it's not about 50% full increase it.

Bob

[Eberhard Wacker]

Raise it to 4096, as mentionend before it's really not much amount of memory even in your case of a small VAX system.
Cheers,
EW

[John Gillings]

The big question here is why do you need more JTQUOTA? What's being defined in the JOB table? Generally speaking, the JOB table is only needed for logical names that must be shared by all processes in the same job tree. Most other logical names can be defined in the process table, which is much bigger.

Are those 150 processes independent, or are they a in a single job tree? This makes a BIG difference. If they're independent then each will have its own JOB table, so whatever you increase JTQUOTA by will be multiplied by 150. If they're in the same job tree the JOB table is shared.

If the increased space is to store many logical names that are common to all 150 (independent) processes, maybe you could define them in the GROUP table instead? (this would also save you from defining them all each time a process starts as they'll be permanent definitions) Another possible alternative would be to create a permanent, shareable logical name table for that user and connect it into a process specific LNM$FILE_DEV search list. This may reduce your memory requirements.

[Sk Noorul Hassan]

Hi,

Free paged pool is 7083872 out of total 8763392.

We are getting logical table full error, when a report generation process tries to allocate it's to be created dettached process name , mail box number/name, filename, logical printer channel to a perticular pool(as per the code). At this point the report generation process comes with SYSTEM-W-EXLNMQUOTA - logical table full error on console.

As per help/mess in VMS, it is suggesting to increase JTquota of the account.

Pls suggest..

[Volker Halle · Accepted solution]

Increasing JTQUOTA for this account should be no problem at all.

You seem to only have ONE application (report generator), which seems to want to create more JOB logical names than the JTQUOTA for this account allows and is therefore failing with EXLNMQUOTA.

If you increase JTQUOTA for this account, this application should then be able to create all of it's necessary logical name entries in the job table.

OTHER PROCESSES running under this account would NOT create more logical names by themselves, if you just increse JTQUOTA, so there will be no additional paged pool resources used.

With increasing QUOTAS, you do NOT increase resource usage directly, you just ALLOW more resource to be used, if required. If other processes don't fail with EXLNMQUOTA, they don't use more than JTQUOTA now and won't be expected to use more of this quota afterwards.

You can check the existing QUOTA usage of all the JOB logical name tables with $ SHOW LOG/FULL/TABLE=LNM$JOB_*

Quota is shown as [Quota=(n,m)]

n = remaining quota
m = quota parameter value (JTQUOTA)

So you could add up all m-n differences for all your processes in this account system-wide. This would tell you the no. of bytes of paged pool used for the logical names in those JOB tables. There is NO REASON to believe, that those bytes used will INCREASE by just increasing JTQUOTA !

Example:

(LNM$JOB_81EA6100) [kernel] [shareable] [Quota=(960,1024)]
[Protection=(RWCD,RWCD,,)] [Owner=[SYSTEM]]

"SMI$NETMBX" [user] = "MBA27:" [terminal]

The hex address value in the logical table name is the system space address of the JIB (Job Information Block).

Volker.

[Sk Noorul Hassan]

Volker,

we have almost 150 processes running(scada APPLICATION), BUT TWO TIMES WE HAVE GOT LOGICAL TABLE FULL ERROR ON THE REPORT GENERATION PROCESS ONLY...FROM NO OTHER PROCESS AS OF NOW.

CHEERS...

[Volker Halle]

So go ahead and change JTQUOTA from 1024 to 2048 for the application account. This will not increase your paged pool utilization or any other resource usage on your OpenVMS VAX system.

Good luck,

Volker.

[Antoniov.]

I hint you change JTQUOTA from 1024 to 2048. For my experience this operation is useful without other trouble.

Antonio Vigliotti