parker@mprgate.mpr.ca (Ross Parker) (10/23/90)
Hi.... I have a periodic problem on a number of Microvaxes running Ultrix 3.1 and 2.3. It appears that the system accounting is shutting itself down for reasons other than disk space. The symptoms are as follows (seen one day on our Microvax-3600 running 3.1): At about 1:00 pm on a weekday with about 15 users on the system: - /usr/adm/acct is approx 100K in size. - df /usr reports that /usr is approx 85% full. - sa -m reports that only 4 users had run any commands. Of these 4, root and daemon used by far the most time. This was a normal day, and there had been at least 15 users on since 9:00 am. Then, at about 2:30 pm, on the same system: - /usr/adm/acct is approx 450K in size. - df /usr still reports approx 85% full. - sa -m reports that a whole bunch of users had run commands. We run a nightly accounting run that clears out /usr/adm/acct and starts it from scratch (we run accton first to turn off accounting, copy and zero the file, then run accton /usr/adm/acct to restart accounting). This nightly accounting run takes about 2 minutes and is run at midnight. As mentioned, this had been a normal day, with lots of activity. Having only 100K worth of accounting data (and only 4 users shown) by 1:00 pm is certainly incorrect. Some time in the next 1-1/2 hours, something happened to restart accounting, as the file grew by a reasonable amount in that time. I unfortunately neglected to check the error log. This problem is happening on about 6 different systems (I'm inferring this from the size of the accounting files on the other systems, and the results of the daily accounting runs - the above incident is the only one I've 'caught'). It's not a disk space problem. Does anyone have any ideas? Is there any reason that accounting will shut down other than disk space limitations? Dec has had this problem for about 3 weeks without results. -- Ross Parker | Why do they put me down? | Make out that I'm a clown? parker@mprgate.mpr.ca | I drink scotch whisky all day long uunet!ubc-cs!mprgate!parker | Yeah I'm gonna save my money | (gonna put it all away...) (604)293-5495 | 'Cause I'm a Scotsman
grr@cbmvax.commodore.com (George Robbins) (10/23/90)
In article <2392@kiwi.mpr.ca> parker@mprgate.mpr.ca (Ross Parker) writes: > > I have a periodic problem on a number of Microvaxes running Ultrix > 3.1 and 2.3. It appears that the system accounting is shutting itself > down for reasons other than disk space. The symptoms are as follows > (seen one day on our Microvax-3600 running 3.1): Are you getting the "accounting shutting down" messages on the console, or does it just seem to stop recording? Your disk space analysis seems reaonsable. If the problem is frequent, you my put something in crontab the logs accout file size and free space on a periodic basis and see if there's any pattern to this disease. -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing: domain: grr@cbmvax.commodore.com Commodore, Engineering Department phone: 215-431-9349 (only by moonlite)
parker@mprgate.mpr.ca (Ross Parker) (10/25/90)
In article <15340@cbmvax.commodore.com>, grr@cbmvax.commodore.com (George Robbins) writes: |> In article <2392@kiwi.mpr.ca> parker@mprgate.mpr.ca (Ross Parker) writes: |> > |> > I have a periodic problem on a number of Microvaxes running Ultrix |> > 3.1 and 2.3. It appears that the system accounting is shutting itself |> > down for reasons other than disk space. |> |> Are you getting the "accounting shutting down" messages on the console, |> or does it just seem to stop recording? I unfortunately haven't been checking the error log (sheepish grin). I know that I periodically do have accounting shut down for disk space reasons, and unfortunately can't remember what day I caught the problem I described, so I can't look back to see if this was registered in the error log. |> If the problem is frequent, you my put something in crontab |> the logs accout file size and free space on a periodic basis and see if |> there's any pattern to this disease. I've done this already... I run a script every 15 minutes that checks disk space, the size of the accounting file, and the kernel variable 'acctp'. If acctp changes (it's a pointer to an internal accounting structure), it means that accounting has either shut down or started up in the intervening 15 minutes. I check also to see if the size of the accounting file remains constant for 15 minutes (highly unlikely when things are normal). I mail myself if acctp changes or if /usr/adm/acct has remained constant. To date, the only times I've received alarms are: - disk space problems - nightly accounting run which turns off and restarts accounting. This script has been going for about a week. I think this is a case of 'the watched pot never boils', as everything seems normal. Grrrrr..... The problem *used* to be frequent!!! -- Ross Parker | Why do they put me down? | Make out that I'm a clown? parker@mprgate.mpr.ca | I drink scotch whisky all day long uunet!ubc-cs!mprgate!parker | Yeah I'm gonna save my money | (gonna put it all away...) (604)293-5495 | 'Cause I'm a Scotsman