Understanding...

solarix · Post by **solarix** » Sat Dec 01, 2001 11:28 am

Just when I thought I was starting to understand.

The ping Time graphs do not draw, I just get a broken graphic icon.

the rrd files are being created, when I access the graph directly using
http://server.daphnis.com/cacti/graph_i ... =8&rraid=4
it returns a 0
access a graph
http://server.daphnis.com/cacti/graph_i ... egend=true

and the graphing functions are working.

[ This Message was edited by: solarix on 2001-12-01 11:42 ]

[ This Message was edited by: solarix on 2001-12-01 11:42 ]

Post by **raX** » Sat Dec 01, 2001 12:02 pm

The problem is with the graph itself. Your first data source is a STACK. You must always use an AREA first, then STACK.

FYI: the "show graph info" button is very helpful in these situations as it lets you look at the rrdtool source and see what is acually being sent to rrdtool.

This bug is my fault though, I forgot to change the STACK to AREA when making the sample graphs. It has been fixed for 0.6.1 however.

-Ian

[ This Message was edited by: raX on 2001-12-01 12:05 ]

solarix · Post by **solarix** » Sun Dec 02, 2001 7:54 am

OK... I am feeling stupid here.

Now all the graphs draw, but there is no historical data being stored for anything. It is as if the rrd's are being recreated from scratch every time?

On occasion the data shows on the graphs, then as I refresh the page the data slowly disappears.

I did some investigation and the rra's are not being recreated. It must be in the graph setup paramaters?

Maybe this makes some sense?
The log file says
12/02/2001 8:05 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/system_webhits.rrd N:66777
12/02/2001 8:05 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/system_mem_free.rrd N:50160
12/02/2001 8:05 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/system_mem_buffers.rrd N:159052
12/02/2001 8:05 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/system_mem_cached.rrd N:76800
12/02/2001 8:05 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/system_mem_swap.rrd N:284612
12/02/2001 8:05 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/system_mem_total.rrd N:354868
12/02/2001 8:06 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/ping_email.rrd N:1.127
12/02/2001 8:08 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/ping_wdaphnis.rrd N:9.361
12/02/2001 8:09 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/ping_gering.rrd N:1.789
12/02/2001 8:10 AM - CMD: /opt/rrd/bin/rrdtool update /www/htdocs/server/cacti/rra/ping_waulick.rrd N:14.185

The minutes are wrong. these updates occured within seconds.

[ This Message was edited by: solarix on 2001-12-02 09:07 ]

Post by **raX** » Sun Dec 02, 2001 11:32 am

I found what the problem is, but I am unsure about why this may be happening.

The problem is with this line in your rrd source:

--start="-144"

For a daily graph, that number should be -86400 (the number of seconds in a day). That would explain why your timescale is way off.

This number is calculated using the following formula for the given RRA:

number = rows * steps * 144

By default cacti comes with all of the nessesary numbers in its database for showing daily, weekly, monthly, and yearly graphs (this can be changed however under Round Robin Archives)

My guess is that your round robin archive information is not correct for some reason or another.

If you could do me a favor, run a:

select * from rrd_rra

...and post it here, hopefully I will be able to catch the problem. I can tell you though I was unable to reproduce this bug using 0.6 on my machine...strange.

-Ian

solarix · Post by **solarix** » Sun Dec 02, 2001 12:24 pm

mysql> select * from rrd_rra;
+----+-------------------+--------------+-------+------+
| ID | Name | XFilesFactor | Steps | Rows |
+----+-------------------+--------------+-------+------+
| 1 | Daily (Average) | 0.5000 | 1 | 1 |
| 2 | Weekly (Average) | 0.5000 | 6 | 1 |
| 4 | Yearly (Average) | 0.5000 | 288 | 1 |
| 3 | Monthly (Average) | 0.5000 | 24 | 1 |
+----+-------------------+--------------+-------+------+
4 rows in set (0.00 sec)

Post by **raX** » Sun Dec 02, 2001 12:36 pm

Wow, the steps/rows for all of those rra's are off. Run these queries to fix it:

REPLACE INTO rrd_rra VALUES (1,'Daily (Average)',0.5,1,600);
REPLACE INTO rrd_rra VALUES (2,'Weekly (Average)',0.5,6,700);
REPLACE INTO rrd_rra VALUES (4,'Yearly (Average)',0.5,288,797);
REPLACE INTO rrd_rra VALUES (3,'Monthly (Average)',0.5,24,775);

See if that helps. Also, since those values were messed up, chances are it would affect the .rra files as well. You may want to delete all of the files in your rra/ directory and let them recreate themselves after you run that query.

I pulled those queries right from my SQL scripts, so my origional values are correct, they must have gotten changed sometime afterwards.

-Ian

solarix · Post by **solarix** » Sun Dec 02, 2001 12:50 pm

Interesting, every time I edit the rra's it changes them to the wrong values...

Now it shows
--start="-33053184"

Maybe I'll wait till 061 to try this and rip out everything and start over...

Looks like "$rows" is used twice or is a reserved word in my versions of mysql or php

[ This Message was edited by: solarix on 2001-12-02 13:06 ]

Post by **raX** » Sun Dec 02, 2001 1:11 pm

Your right, there was a bug that was not allowing the "row" column to be saved correctly. I'm glad you caught that one, it has been fixed for 0.6.1 however.

Acually $rows is not a reserved word in PHP, but a variable that I use to represent how many rows were pulled out in a SQL query... No wonder it was always '1'.

Hey on another note, I noticed that you were monitering ping times to 'sol.intertech.net'. What sort of affiliation do you have with them? My ISP used to have satelite service through them.

-Ian

[ This Message was edited by: raX on 2001-12-02 13:20 ]

solarix · Post by **solarix** » Sun Dec 02, 2001 1:26 pm

Glad it's a bug! I was going crazy!

While we're bug slapping... the graph_image.php sends type text/html when it is really a png/gif... maybe not a problem but it is there none the less.

part of our bandwidth comes through them... they have a bottleneck I am wanting to point out

[ This Message was edited by: solarix on 2001-12-02 13:40 ]

Post by **raX** » Sun Dec 02, 2001 4:50 pm

I guess I should have known that 'Type' is not a valid HTTP header type. I fixed it however, and changed it to 'Content-type'.

When my ISP used Intertech about a year ago, our service was terrible. They would disconnect users for over-downloading, while it seemed they were running ~14 cable ISP's over two T1's...ouch.

I happened to notice your MySQL graph... 40 queries/second, that is quite a busy server you are graphing it sounds. I only wish I was as priviliged.

Also, would you mind if I use your graph page on cacti's list of sites? I am hoping to gather a list of people using cacti to show the many things it can be used for.

-Ian

[ This Message was edited by: raX on 2001-12-02 19:54 ]

solarix · Post by **solarix** » Sun Dec 02, 2001 8:34 pm

I will be monitoring some items I have non disclosure agreements on... they will not be open to the public, even if I have to .htaccess cacti.

Is there a way to secure areas and not others?

Cacti

Understanding...

Who is online