12-28-2010 05:05 PM
Im trying to set up an event where i get an email notification when a port exceeds certain CRCs or link failures. User manual is kinda too general for this and I can't figure out how to do it
Any help appreciated
01-03-2011 03:46 AM
do you have a technical issue with generating messages, sending alerts or policys?
Or you you not know what to monitor?
Are you looking for specific counter thresholds?
In the last case it is difficult to give a general answer because every environment is different. You have to find your own settings.Check the FabricWatch Admin guide. There you can find some ideas how top monitor your fabric.
I hope this helps.
01-07-2011 04:12 PM
Ok let me be a little more specific.
Version is CMDCE 10.1.1 (connectrix manager)
Test e-mail works
When I create an event policy, whatever change I make (like zoning or port changes etc) the event is not triggered.
I know the switch has registered its SNMP traps.
In the master log, I do not see these changes I make (here is where i thing the problem is)
I have no filters configured
Hope you can help
On another note, you know if the new BNA also manages SANs? what about licensing?
01-09-2011 11:10 PM
sorry for the delay.
I am not sure if your version 10.1. connectrix manager is the same as DCFM 10.1 if so I would suggest to update to 10.4 latest release. '
In the old 10.1 and 10.2 DCFMs are many unexpected features ;-)
In our environment it was not stable.
BNA manages SAN and IP Networks.
Currently the SAN part does not have the same functions as DCFM10.4
Brocade provides a trial version of 75 days.
If you have a vaild 10.3 or 10.4 license of DCFM you move onto BNA 11 without any additional costs. I do not know if EMC have the same license model.
Coming back to your issue.
One basic question, can you see on the masterlog the same messages as on the san switches?
What I want to say is if your DCFM masterlog does not see the messages it will not be able run any policies.
The switches will send all messages via syslog to the DCFM so you have the need to run on DCFM the integrated syslogserver and the switches should send these messages to DCFM. If your DCFM server has an alternate syslogserver running it will not work correctly.
Please check with syslogfipshow if the DCFM ip is included.
01-11-2011 02:20 AM
I believe i have a very similar scenario to Roberto on an Enterprise DCFM 10.4.4
I would appreciate some more detail on Andreas last comment on the use of the syslogipshow command - is this a FOS shell command that
we need to run on the switches?
I have wondered why the master log on our DCFM installation seemed devoid of any interesting activity - scant information about any port changes or ISLs.
I also would find a detailed checklist on syslog configuration useful - or a pointer towards the relevant documentation which i may have overlooked.
I have found getting any relevant email alerts out of DCFM much more difficult that i had expected. If anyone knows of a useful general document please
provide a link.
01-11-2011 02:38 AM
connect to the CLI of a san switch with telnet or ssh.
run the command syslogdipshow and check if your DCFM server is included in the list.
If not run syslogdipadd ww.xx.yy.zz command to add your DCFM server IP.
In addition check if SNMPv1 or snmpv3 in case of Virtual Fabric is configured correctly.
Make sure that your DCFM server can run snmp reads from the san switch. On linux you can use the installed snmpget commands. For Windows you get get some free tools in the internet.
If you wolud like to monitor any changes run the CLI command trackchangesset 1,1
This enable to monitor: login, logouts, config changes and track-changes on / off changes.
The last option is to hire a Brocade professional.
I hope this helps.
01-11-2011 02:49 AM
First of all let me say that i already figured it out, and since all documentation regarding this is so vague, here's the procedure I found to work.
First of all, version 10.1.1 does have problems... so dont use that one, the one im using now and working is 10.4.3
This is taking into account you're running a clean install of DCFM, with default syslog installation and ports. (no need to run that syslog command)
You have to actually enable snmp traps on each switch so they can appear on the master log, the command i believe is "snmpconfig --set snmpv1"
In my case, we have DCX, 48ks and 24ks, none of em have them enabled by default
You have to make sure the severity level is set to 4 (snmpconfig --set seclevel), and enable traps for SW-TRAP options, including swFCPortScn,
swEventTrap, swFabricWatchTrap and swTrackChangesTrap. (with the first command I specified)
In my case i do not need FA-TRAP (not even sure what exactly it "traps") but you can enable it, or HA-TRAP even.
Once you've done this, delete the switch or fabric from the discover option and redescover it, so DCFM can reconfigure it.
With this you're all set to actually see the event logs from the switch in the master log, which is why the event notifications didnt work in the first place.
From here its pretty easy to just try different event policies. I think you can only have 10.
Also, check the user options and its filter for e-mail notifications, I had to re-do my user database in order to just leave certain filters on or else everyone will just get email spam from even INFO types of events.
Good Luck!, hope this helps everyone since I could not, for the life of me, find any documentation AT ALL for this procedure. (yeah, not even direct support from EMC)
Edit: I just reread this, the exact command to change it is: "snmpconfig --set mibcapability" more info on the attached pdf
01-11-2011 02:51 AM
I forgot to mention,
although i can monitor everything with this now, i can't still manage to monitor high CRC error or link failure counts. Port fencing is not good enough for me, since my interest is not in Shutting down the port, rather smartly prevent them from going bad on production hours (aka, replace them on time before they fail).
Any ideas on this?
01-19-2011 09:10 AM
For your problem, do you have the Fabric Watch license deployed?
Anyhow I have a question of my own
I try to trigger a perl script with an event policy.
DCFM doesn't seem to be able to launch it with the following message;
Failed: Unable to execute the Action Launch Script for the Event Policy Array Port Offline : Error while launching a script
The script runs well when I launch it separatly. I'm thinking of an environment variable issue but DCFM seems to be working with "local system" account.
Any idea on how to check the env for this special user?
01-19-2011 03:03 PM
I do have licensing for Fabric Watch, however I never got the chance to really go deep into it, why the question?
Regarding your script, I'm not too perl savvy, but is there a way to make it launch as a global user? within the script itself?