| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1585.2 | Raise PGFLQUO | AIMTEC::VOLLER_I | Gordon (T) Gopher for President | Fri Oct 09 1992 00:00 | 14 |
| Randy,
The ALLIN1 MANAGER's account is running out of VM.
The reason for this is PGFLQUO (100,000 - from UAF) being
exceeded as the amount of VM in use is 116,964 (Peak virtual size -
from Accounting information.
Solution = raise ALLIN1's PGFLQUO (try 200,000 - maximum allowed
due to VIRTUALPAGECNT value)
Cheers,
Iain.
|
| 1585.3 | WSEXT | IOSG::TALLETT | Gimmee an Alpha colour notebook... | Fri Oct 09 1992 09:10 | 16 |
|
...and for performance improvements, I think I'd raise
the manager's WSEXTENT as the FCS is banging up on the
4096 working set limit and page faulting (admittedly
probably only soft faults).
Not that I would even hint at the suggestion of a bug, but
you might want to watch the FCS's Virtual pages from time
to time for a memory leak, as 100,000 sounds a bit excessive,
ours only uses 20K pages.
How long had this FCS been running? Were there a lot of active
users at the time?
Regards,
Paul
|
| 1585.4 | Still upgrading - no user load yet | DV780::BAILEYR | OICU812! | Fri Oct 09 1992 16:53 | 23 |
| Iain/Paul,
>How long had this FCS been running? Were there a lot of active
>users at the time?
We're still in the process of ugrading from 2.4. We had profile
problems which caused the A1$POSTINSTALL grief, so we're
performing each step manually now. The only user logged in was
the manager and a system-type who was just monitoring the manager
process. The FCS was up for approx. 24 hours prior to the
DO OAFC$PART_SEED attempt and again - no activity.
I bumped up the PGFLQUOTA to 200k and increased the WSEXTENT to
7680 and will advise the results. The system in question is an
8250 with 20mb of memory and is being used as a test system to
allow us to upgrade and migrate any 2.4 customizations off-line.
STARS mentioned *************************** related to memory
problems. Could this be related to our problem?
Tnks again!
Randy
|
| 1585.6 | | IOSG::PYE | Graham - ALL-IN-1 Sorcerer's Apprentice | Fri Oct 09 1992 18:41 | 11 |
| I've hacked the previous note, because it mentioned a possible patch
which is not committed, or available and we prefer not to discuss
futures in this notesfile.
I've not heard of a problem that would explain the effect you're
seeing, and considering how lightly your system is loaded, it seems to
be taking a damm long time to do the seeding!
Perhaps Paul's right and you've discovered a new memory leak???
Graham
|
| 1585.7 | Raising WSEXT helped! | DV780::BAILEYR | OICU812! | Fri Oct 09 1992 20:45 | 23 |
| Looks like raising WSEXTENT certainly helped! After I raised it
to 7680 and started the OAFC$PART_SEED script again, I watched it
from another process and "Peak ws size" was right around 4310 for
over an hour and "Page faults" were down around 5500 or so.
Sometime during approx. the last 30 minutes the peak ws hit 7680
and the number of pg faults reached 29750 or so, but still
allowed the script to complete.
Buffered I/O count: 88868 Peak working set size: 7680
Direct I/O count: 24904 Peak virtual size: 41196
Page faults: 30507 Mounted volumes: 0
Images activated: 7
Elapsed CPU time: 0 01:16:08.50
Connect time: 0 03:33:47.13
>I've hacked the previous note, because it mentioned a possible patch
>which is not committed, or available and we prefer not to discuss
>futures in this notesfile.
Oooppps! Just trying to keep Graham on his toes ;) My apologies...
Thanks again for the aid!
Randy
|
| 1585.8 | Doesn't look like a leak anymore | IOSG::TALLETT | Gimmee an Alpha colour notebook... | Mon Oct 12 1992 09:24 | 10 |
|
I take back my suspicion of a memory leak. It sounded as if the
FCS had been running for a long time and then keeled over, but
if you're still doing the upgrade, then the FCS hasn't had much
to do at all - it is much more likely to be system setup problems.
Okay JD, I'm ready for the punishment....
Regards,
Paul
|