ZenPacks : Amazon Web Services : Comments

RE: Amazon Web Services

community@zenoss.org — Tue, 02 Feb 2010 23:08:35 GMT

To monitor a Linux EC2 instance via SSH:

Install the Linux Monitor ZenPack.
Save your public key from the AWS site on your monitoring server (ie. zenoss.pem)
Add your EC2 instance under Add Device
1. Device Name is the instance's Public DNS
2. Device Class Path is /Server/SSH/Linux
3. Set the Discovery Protocol to "None"
Go to the new Device.
Under zProperties set the zKeyPath to point to it (ie. ~/zenoss.pem)
Under zProperties set the zCommandUsername (ie. "ubuntu" or "root")
Under zProperties set the zPingMonitorIgnore to "True".
Model the device.

You could use one of the more specific Linux ZenPacks if necessary (or other OS as applicable).

RE: Amazon Web Services

community@zenoss.org — Wed, 14 Jul 2010 00:18:24 GMT

I can only see East instances. Is there any patch (http://dev.zenoss.org/trac/ticket/5811) for 2.5.X Zenpack?

RE: Amazon Web Services

community@zenoss.org — Thu, 15 Jul 2010 16:03:04 GMT

This is fixed in the Zenoss 3.0 version, I assume you could compile this to run on Zenoss 2.5.2 and it still would work.

RE: Amazon Web Services

community@zenoss.org — Tue, 21 Dec 2010 11:13:34 GMT

Isn't fixed on 3.0.3, we still having the same problem, the zone is hardcoded on connection.py and anyway the graphs doesn't appear after changing manually the zone, seems nobody from Zenoss wants to see this problem solved? too many months, too many forum messages but nobody solved this. I thinks that is an important zenpack for all users so I don't understand how could be wrong after a lot of messages asking for the solution.

RE: Amazon Web Services

community@zenoss.org — Fri, 25 Feb 2011 05:03:13 GMT

Do we have a way to disable automatic activation of Cloudwatch?

1) AWS API provides an excelent discovery mechanism without having to port scan.

2) CloudWatch stats are not needed when SNMP can be used.

3) CloudWatch costs money and delivers no features beyond what SNMP can provide.

4) It's logical to assume that many would want to run the discovery portion without the unneeded cost associated with CloudWatch.

RE: Amazon Web Services

community@zenoss.org — Wed, 09 Mar 2011 21:26:22 GMT

I believe that the issue with CloudWatch Detailed monitoring is an issue with how Amazon is handling CloudWatch API calls.

So what I am saying is this - If you use their API to grab performance data then the monitoring type is auto-magically flipped from Basic to Detailed. If thats the case, then this is not an issue with Zenoss or whatever monitoring solution you use that leverages CloudWatch API.

RE: Amazon Web Services

community@zenoss.org — Thu, 10 Mar 2011 02:26:51 GMT

"whatever monitoring solution you use that leverages CloudWatch API" = um............ This ZenPack...................

If I switch off all CW based stuff in the templates the code still switches on Detailed stats even when not required in the slightest.

The ZenPack needs only to make a call to list the Instances from the standard APIs.

However it is turning on detailed stats. For no reason.

I "fix" this by running a cron every 15mins to disable monitoring on instances.

Yes you are correct, the problem is the way this AWS ZenPack functions.

RE: Amazon Web Services

community@zenoss.org — Wed, 14 Sep 2011 17:14:49 GMT

I am see error messages in zenhub.log every 5 minutes:

2011-09-14 17:56:54,238 ERROR zen.Events: Required event field severity not found -- ignoring event

2011-09-14 17:56:54,238 ERROR zen.Events: Event info: {'firstTime': 1316019414.238245, 'monitor': 'localhost', 'component': 'zencommand', 'agent': 'zencommand', 'summary': 'Error loading parser <Products.DataCollector.Plugins.PluginLoader instance at 0x9f1fdec>', 'manager': 'zenoss.class.example.org', 'eventKey': '', 'device': 'EC2Manager', 'message': 'Traceback (most recent call last):\n File "/usr/local/zenoss/zenoss/Products/ZenRRD/zencommand.py", line 663, in parseResults\n parser = cmd.parser.create()\n File "/usr/local/zenoss/zenoss/Products/DataCollector/Plugins.py", line 111, in create\n traceback=traceback.format_exc().splitlines())\nPluginImportError: (\'Traceback (most recent call last):\', \' File "/usr/local/zenoss/zenoss/Products/DataCollector/Plugins.py", line 102, in create\', \' self.modPath)\', \' File "/usr/local/zenoss/zenoss/Products/DataCollector/Plugins.py", line 173, in importPlugin\', \' return importClass(modPath)\', \' File "/usr/local/zenoss/zenoss/Products/ZenUtils/Utils.py", line 365, in importClass\', \' raise ex\', \'ImportError: No module named zenoss.ZenAWS.parsers.ec2.instances\')\n', 'lastTime': 1316019414.238245}

This is ZenPacks.zenoss.ZenAWS-1.0.3. I have exactly the same on a 3.1 and a 3.2 system. Does anyone else see them???

Update - the end of this error message is correct - there is no module named zenoss.ZenAWS.parsers.ec2.instances - it should be ZenPacks.zenoss.ZenAWS.parsers.ec2.instances .

What's going on?

Cheers,

Jane

Cheers,

Jane

RE: Amazon Web Services

community@zenoss.org — Wed, 14 Sep 2011 17:46:56 GMT

Jane, what are you trying to setup and what is the functional issue?

Ex: you are trying to monitor 5 test ec2 instances and you can see the instances in zenoss UI, but no stats. Or maybe you see no instances show up in the UI...

Also, you arent trying to monitor multiple AWS accounts right? You can only use the one default EC2Manager with one login.

To me, the error seems to point at a bad install or configuration of the zenpack - it says it cant find the parser module, which comes installed with the ZenPack. You might try uninstalling, looking for remnants, and reinstalling. Also, you cant move around or rename device classes for it, or the 'EC2Manager'. Its kinda hard wired in place.

RE: Amazon Web Services

community@zenoss.org — Thu, 15 Sep 2011 12:41:19 GMT

The discovery works fine. I have configured with the appropriate keys and have set default device classes (/Server/Linux and /Server/Windows) which seem to work fine. Instances doscovered. The DNS names are linked to the instances under EC2Manager. I haven't changed that name at all. It's a very basic setup. No data under graphs for EC2Manager and no files for it under perf/Devices - cos it can't apparently find the parser. I have checked and the file is actually there.

Now have exactly the same on 3 different systems.

Any ideas?

Cheers,

Jane