| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1519.1 | | CSC32::BUTTERWORTH | Gun Control is a steady hand. | Tue Apr 15 1997 10:10 | 40 |
| > I have two problems with PCM 1.6-400.
>
> 1) on an Alpha running VMS 6.2 we had a storm of NCL events last week,
> filling up the pcm log disk. To fix this temporarily I lowered the
> time in my scheduled archive job, from 7 to 3 days. But the disk was
> just as full. I couldn't understand why, until I made a dir/size=all.
>
> Many logfiles had size like e.g. 47875/356200, that is they still had
> allocated the same diskspace as before the archiving. The only way I
> could find to fix this was to copy all the files to another disk, then
> they got a correct size, like 47875/47920. Deleted original files and
> copied back and restarted pcm. Now it looks OK. I haven't seen any
> behaviour like this before, could this be some change introduced by
> ECO4? When a logfile is archived, surely it's allocated space must be
> equally decreased?
The way it's supposed to work is the file should be truncated and it
certainly used to work that way. Log an IPMT.
> 2) A new install of 1.6-400 ("upgrade" from vcs) on a VAX/VMS 6.1
> system. Archives wont work, whether run scheduled or interactively we
> get this for all systems, even if logfile is very small, so diskspace
> can't be the problem.
> Starting archive procedure for system copcmt
> Time file Pass 1: Working... done.
> Time file Pass 2: Working... done.
> Log file Pass 1: Working... done.
> Log file Pass 2: Working... done.
> Archive procedure for system copcmt failed
>
Usually when I see this it's because there is insufficient space for
the work-file that archive needs and this usually happens becuase the
archive area and the active logging area are ont he same disk. Is that
the case here?
Regards,
Dan
|
| 1519.2 | | SWETSC::WESTERBACK | Panta rei | Wed Apr 16 1997 08:49 | 73 |
| Re problem 1, I'll watch the logfiles for a couple of days to see if
the same scenario reappears, if so I'll IPMT
Re problem 2, doesn't seem to be disk space problem. Console$root and
logfiles are on one disk with 170000 blocks free, archive on another
with 446000 blocks free. (These are RF72's at ~2 mill blocks).
I try to manually archive COPCMT:
COPEMA >> dir/date/size=all console$logfiles:copcmt*
Directory CONSOLE$ROOT:[LOG]
COPCMT.EVENTS;1 0/3 3-MAR-1997 13:15:28.04
COPCMT.LOG;1 1588/1812 3-MAR-1997 13:15:28.15
COPCMT.TIMES;1 307/402 3-MAR-1997 13:15:27.88
Total of 3 files, 1895/2217 blocks.
COPEMA >> console archive/keep=96/confirm copcmt
POLYCENTER Console Manager
Archive facility Version V1.6-400
Copyright (C) 1996, Computer Associates International, Inc. All Rights
Reserved
Archive all except the most recent 96 hours of data for COPCMT system?
(Y/N) [N]: y
Starting archive procedure for system copcmt
Time file Pass 1: Working... done.
Time file Pass 2: Working... done.
Log file Pass 1: Working... done.
Log file Pass 2: Working... done.
Archive procedure for system copcmt failed
At the same time from another session I'm doing
COPEMA >> dir/date/size=all console$archive:copcmt*/mod/since=yes
I can see how the old and new .times, then .log files are being
processed, then the last one before the job fails says:
Directory DISK$DISK1:[CONSOLE.ARCHIVE]
COPCMT_NEW.EVENTS;1
0/0 16-APR-1997 16:37:26.97 [SYSTEM]
COPCMT_NEW.LOG;1 587/588 16-APR-1997 16:37:27.09 [SYSTEM]
COPCMT_NEW.TIMES;1 115/117 16-APR-1997 16:37:26.85 [SYSTEM]
COPCMT_OLD.EVENTS;1
0/0 16-APR-1997 16:37:26.62 [SYSTEM]
COPCMT_OLD.LOG;1 1002/1002 16-APR-1997 16:37:26.73 [SYSTEM]
COPCMT_OLD.TIMES;1 no such file
Total of 6 files, 1704/1707 blocks.
If I look at the event files in the log directory, they are all of size
0/3 or 0/501, expect some 1/3. In my other system they are like 16/264
Can there be something wrong with the event files? On the other hand
the archiving job did work for the first 10 days or so since setup.
Any ideas?
Hans
|
| 1519.3 | | CSC32::BUTTERWORTH | Gun Control is a steady hand. | Wed Apr 16 1997 10:31 | 6 |
| Hmmm. Can you make the EVENTS files available on the E-net for
copying. I would like to have them for at least 2 nodes that are
failing.
Regards,
Dan
|
| 1519.4 | | SWETSC::WESTERBACK | Panta rei | Thu Apr 17 1997 02:22 | 25 |
| I've put some event files on the decnet default directory.
Note that all systems fail to archive.
I just realized that COPEMA was rebooted on April 6, and it seems
that archiving worked after that, until April 15 which was the first
time it tried to archive data logged after the reboot. It seems like
data logged after the reboot fails to archive. As it's not very often
we install PCM, can we have missed something in the configuration,
which needs to be in the startup?
ALPHON >> dir/date/size=all copema""::*.events
Directory COPEMA""::SYS$SYSDEVICE:[FAL$SERVER]
CONSOLEMANAGER.EVENTS;1
469/471 29-JAN-1997 08:45:52.00
COPCMT.EVENTS;1 0/0 3-MAR-1997 13:15:28.00
CSO.EVENTS;1 1/3 1-APR-1997 16:28:56.00
TSP01.EVENTS;1 0/0 1-APR-1997 16:28:49.00
Total of 4 files, 470/474 blocks.
Rgds,
Hans
|
| 1519.5 | | SWETSC::WESTERBACK | Panta rei | Thu Apr 17 1997 02:58 | 11 |
| Addition to .4:
A new reboot makes no change.
Just a note re the new installation, FYI even if it's not relevant:
When we installed, and wanted to define console$archive according
to the manual, there was no CONSOLE$PRIVATE_SETUP.TEMPLATE supplied.
We had to copy it over from the other system running PCM.
Hans
|
| 1519.6 | | CSC32::BUTTERWORTH | Gun Control is a steady hand. | Fri Apr 18 1997 12:37 | 26 |
| Hans,
You have done nothing wrong.
I analyzed your files and found:
$DIR/SIZ *.EVENTS
Directory DUMPS:[BUTTERWORTH]
COPCMT.EVENTS;1 0
CSO.EVENTS;1 1
TSP01.EVENTS;1 0
Note that COPCMT and TSP01 do not have any events in them. Now thats
okay as a node doesn't have to have triggered any events.
The code isn't doing to right thing is seems with empty events files.
Notice that CSO.EVENTS;1 is only using one block (out of 3). There were
only two events: One Connect event and One Disconnect event. I would
imagine that archive isn't doing the right thing with small events
files either. Chances are the problem is the same section of code.
Plz IPMT this. Be sure and keep the files as they may be needed.
Regards,
Dan
|