Loss of capturing alarm frames for hours then fine

Support and queries relating to all previous versions of ZoneMinder
Locked
john_l_b
Posts: 37
Joined: Thu Aug 02, 2007 2:20 am

Loss of capturing alarm frames for hours then fine

Post by john_l_b »

This is an openSUSE 10.3 system with ZM v 1.23.3. For the last several months, it has recored the frames properly so they can be viewed. Recently it started having 'black outs' where the image live is fine, alarms are triggered, a numbered directory created, but no frames get stored. This will go on for several hours, then start properly recording.
It also will do it on one or two of the 5 cameras, and the rest fine, and may record all properly for a while.

The events with zero frames will never display if clicked on, while the others do.
Zones have not changed, and I do keep this cleaned out to about 60% of the hard drive in use. The cleaning is done by clicking the 'all' square, letting it select all displayed event lines, and then clicking [delete]. This is done with the oldest events.

Clicking on 'events' lists several numbered alarms, as:
Id Name Monitor Cause Time Duration Frames Alarm
Frames Total
Score Avg.
Score Max.
Score
175769 Event-175769 Trash Continuous 01/06 12:40:00 599.77 1142 0 0 0 0
175773 Event-175773 Trash Continuous 01/06 12:50:00 600.27 1139 6 260 43 75
175779 Event-175779 Trash Continuous 01/06 13:00:00 599.76 1134 15 247 16 41
175784 New Event Trash Continuous 01/06 13:10:00 211.14 400 16 283 17 28
175790 New Event Trash Continuous 01/06 13:14:13 0.00 0 0 0
175795 New Event Trash Continuous 01/06 13:14:29 0.00 0 0 0
175800 New Event Trash Continuous 01/06 13:14:49 0.00 0 0 0
User avatar
cordel
Posts: 5210
Joined: Fri Mar 05, 2004 4:47 pm
Location: /USA/Washington/Seattle

Post by cordel »

Okay so obviously not the HD filling up.
What size is the drive and what filesystem are you using on it?
ext3 has a 32k file limit for each folder for example. So if you have 32k events on one monitor, that folder would be at it's limit.
The other thing you might check is your DB health.
john_l_b
Posts: 37
Joined: Thu Aug 02, 2007 2:20 am

Post by john_l_b »

It has ext3, and that file limit could be the problem. The DB could probably use a solid repair or rebuild.

Right now I try to keep it down to one month of events per camera, I'll reduce that down to three weeks and check the results. If it does, I'll change it over to reiserfs when i update to openSUSE 11.1 and ZoneMInder 1.23.x or 1.24.x if it's out of beta then.

Another question:
Some of the pro camera companies have been coming out with high quality DVR systems, in Linux, and they claim that they program them to work with IE as a browser exclusively.
The claim is that it is easier to program for and use ActiveX on IE then Firefox on Linux.

My claim is that is why there are tons of malware for Windows, so I don't want to use Windows unless there is absolutely no other option.
john_l_b
Posts: 37
Joined: Thu Aug 02, 2007 2:20 am

New information

Post by john_l_b »

I went down and looked at the ZM server, and deleted via zm older events, leaving about 2 weeks on each camera, and set cameras to monitor only.

Then I did a 'df' and found the drive was at 100% usage.
One thing strange about that is that it has never added 40% of disk usage in a couple of days.
Now, after a few hours, it is still at 100% usage.
So I have written a script to delete all other old event files that are not in the current database.

I give the results here afterwards
john_l_b
Posts: 37
Joined: Thu Aug 02, 2007 2:20 am

Post by john_l_b »

It's now at 71% usage on the disk. and set to record again
Another strange thing - an event in the database that shows 0's for everything has 737 frames saved. There is also a final entry in the directory that is '.163537' - a period followed by the directory number.

It does look a lot like db corruption.
User avatar
cordel
Posts: 5210
Joined: Fri Mar 05, 2004 4:47 pm
Location: /USA/Washington/Seattle

Post by cordel »

I'm sure it is DB corruption. ext3 reserves something like 10% of the drive for the root processes, this does not include mysql. So once the drive filled up, mysql could not add entries into the database and has become corrupted.
Interesting that the console only shows 40%, there must be a config error some where.
What do you have set for ZM_EVENT_PATH?
Should be set to "events".
coke
Posts: 518
Joined: Wed Jan 30, 2008 5:53 pm
Location: St. Louis, MO, USA

Post by coke »

While you're at it, there's already a script to delete events that aren't in the database. zmaudit.pl

It runs itself in the background normally, but you can invoke it manually.
john_l_b
Posts: 37
Joined: Thu Aug 02, 2007 2:20 am

Now working fine

Post by john_l_b »

After the manual (script based) deletes, it got down to 72% and started recording properly.

I'll check out zmaudit.pl. It apparently was not running in the background, or the system doesn't have enough CPU to handle it. It does 5 cameras with an average of 1.5 fps. No IP cameras yet but some planned.

I did the deletes with zm not running. The system uses a Intel Celeron D at 2.7G and 1G RAM and 4G swap. I'm considering using an AMD Athlon(tm) 64 X2 Dual Core Processor 5000+ with the ASUS M2A-VM motherboard. I have that motherboard and a single core AMD running as a server for the overall network, and one with that dual core really performs well. I just haven't done any ZM on it yet.
Locked