Brocade Fibre Channel templates.

pirx · Posted: Mon Jan 07, 2008 10:01 am Post subject:

Can anynone tell me what I've to do to get the proper graphs with the Bytes/Second template.

During a benchmark the Brocade 200E Switch shows me this numbers (for the time of the benchmark, ~5 min dd read from raw disk)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total
======================================================================================
289m 290m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 580m

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total
======================================================================================
294m 296m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 591m

But the graph shows only ~80-90M.

What do I have to chose when cacti ask me this:

Data Source [Summary of RX & TX Word]
A CDEF (math) function to apply to this item on the graph.

Currently I choose none.

ame · Cacti User Joined: 25 Nov 2005 Posts: 68

choose the "translate brocade words into bytes"-CDEF
It sums the RX and TX words and multiplies it with 4 - so you get the throughput bytes.

Change this in the graph template for the first entries of "throughput bytes" and cacti does not ask anymore....

But when I look to your graphs it seems correct....

Alex

fosforo · Joined: 28 Jun 2007 Posts: 2

pirx · Posted: Wed Jan 09, 2008 3:32 pm Post subject:

ame · Cacti User Joined: 25 Nov 2005 Posts: 68

which values did you monitor on your brocade? bytes/sec? words/sec?.....

We have here many really fast IBM storage boxes but none of them reaches 290 MB/sec.....

So first you should clarify which values you are monitoring on your brocade

Alex

pirx · Posted: Mon Jan 14, 2008 11:39 am Post subject:

ame · Cacti User Joined: 25 Nov 2005 Posts: 68

pirx,

I wanted to know what you have measured at your own benchmark test..

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total
======================================================================================
289m 290m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 580m

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total
======================================================================================
294m 296m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 591m

the values above - are they MB/s, Bytes/sec, words/sec....?

Try to make a test with IO-Meter (free downloadable)....after that you can compare the results much better....

Alex

pirx · Posted: Tue Jan 15, 2008 4:21 am Post subject:

ame · Cacti User Joined: 25 Nov 2005 Posts: 68

pirx...I remember that we had a similar problem with running cacti (0.86) under debian. I thought that our SAN has nothing to do - peaks until 110 MB/s. Then I tried for updating reasons CactiEZ (0.87) installed on a VM and imported my Brocade templates. Switch to 1 minute polling and lay back....

Alex

gandalf · Posted: Wed Jan 16, 2008 4:47 pm Post subject:

trevorg · Joined: 09 Mar 2008 Posts: 2

Brocade uses 32bit counters. So yes you will hit the limit if you have heavy traffic in a 5mins period.

I got around this by using frames * 2112 (2112 is the size of the data portition of the fc frame). The only problem here is the FC spec says the data payload is variable but i have yet to see any issues in my testing.

Here is what i put in mrtg but im working on cacti side of things very soon.

1.3.6.1.4.1.1588.2.1.1.1.6.2.1.14.2&1.3.6.1.4.1.1588.2.1.1.1.6.2.1.13.2:public@x.x.x.x * 2112

gigglesworth · Joined: 25 Jan 2006 Posts: 22

Thanks for these templates.

I notice your templates include a Graph Template called "Interface - Traffic - In/Out and Errors/Discards". Does this graph template intentionally work with regular Ethernet Interfaces?

I accidentally applied this Graph Template to some regular Ethernet Interfaces, and I get actual graphs. However, I'm not sure if the values in these graphs are valid.

I was wondering what the heck the labels "From CPE" and "To CPE" were doing there

.

lozanol · Joined: 19 Jun 2008 Posts: 1

We have 4 GB Brocade Switches and i have also the problem with the 32 bit counters.

I'm a cacti newbie, and i don't know how to change from tx, rxwords x 4 --> rx, txframes x 2112

I have try it, but every time i must reload the template. Can someone help me ?

Luis

tomer · Joined: 26 Aug 2008 Posts: 1

hi all,

it seems that i have exactely the same problem like lozanol. i would like to use cacti to monitor the backup san using cacti. the san contains only 4-gbit san switches.

basically it seems that cati is working with the scripts i found here but i'm pretty sure that the values which are reportet by cacty are false. looking at the graphs i can see that we have a maximum of 80Mbit traffic on the switch ports. if i'm looing at the switches directly i can see much higher valuers. also looking at the backup application i can see that there is much more traffic on the switchports (if some DB backups are running we have up to 350MB/s of data on some swicth ports). i know that cacti onyl displays some average values but the backups jobs are running for hours and i still won't get any values higher than 100Mbit reported.

does anyone have similar problems with monitoruing 4gbit san ports ? is there a known way how a 4gbit san port can be managed ?

thanks for any feedback,
tomer

trevorg · Joined: 09 Mar 2008 Posts: 2

If you dont know how to get around the issue the "simple" way would be just use the existing brocade template. Rip everything out of the FC switch graph template except for RX/TXframes and then place a CDEF on the graph items to perform FRAMES * 2112, which will give you bytes. So far this has worked very well for me.