| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 3808.1 | An error return to Alarms from the Bridge AM | MOLAR::ROBERTS | Keith Roberts - Network Management Applications | Mon Sep 28 1992 10:36 | 22 |
| > The requested operation cannot be
> completed<EOS>%MCC-E-MEM_ALLOC_FAILE,
> memory allocation failure
Dan,
This appears to be an error directly returned to Alarms from the Bridge AM.
Memory usage is very dynamic. It is possible that one command yields
an out-of-memory error, and yet the next command does not.
What were the conditions while this rule was running? Were there a lot
of other rules executing too? Do you have any idea how much memory the
entire process was using at the time?
Use 'show proce/cont/id=<pid-number>' and watch the process virtual page
count. If it is approaching the 'virtualpagcnt' sysgen parameter then
you were indeed running out of memory. Otherwise, ... Can someone from
the bridge team give an indication as to why the Bridge AM would return
this error ?
/keith
|
| 3808.2 | VIRTUALPAGECNT parameter is OK, but lots of rules firing | CUJO::HILL | Dan Hill-Net.Mgt.-Customer Resident | Wed Sep 30 1992 02:04 | 21 |
| Keith,
This occurred on a VAXstation 3100/38 with 32MB memory.
VIRTUALPAGECNT=100000. Process had 18000 page working set out of 32000
possible. Virtual pages in use were 48000. User PGFLQUO=120000.
Alarm rules were enabled in a separate FCL window for testing purposes
instead of via batch. I was not monitoring the FCL "alarms" window, so
it could be that this process was running out of resources, though it
had exactly the same default set-ups as IMPM startup process.
Three global alarm rules were enabled. One did bridge reachability,
one did ipReachability, and the other tested Station entities. 20
bridges were being polled every 5 minutes, as were 7 stations and 200
ip nodes.
I have noticed a few other items that should possibly be increased
based on what I am doing. I'll make the changes and let you know if
the problem persists.
Thanks,
Dan
|
| 3808.3 | Parameters look reasonable | MOLAR::ROBERTS | Keith Roberts - Network Management Applications | Wed Sep 30 1992 10:30 | 23 |
| RE: .2
Dan,
Your parameters look reasonable:
> This occurred on a VAXstation 3100/38 with 32MB memory.
> VIRTUALPAGECNT=100000. Process had 18000 page working set out of 32000
> possible. Virtual pages in use were 48000. User PGFLQUO=120000.
And you don't appear to have 'too many' things going on to eat up all
that memory.
> I have noticed a few other items that should possibly be increased
> based on what I am doing. I'll make the changes and let you know if
> the problem persists.
Great. In the mean time .. can the BRIDGE team give any reasons why
this error (MCC-E-MEM_ALLOC_FAILE) might be returned. This could even
be coming from the Kernel Ethernet Access routines too, I imagine.
/keith
|