I'll answer in-line:
mlist wrote, On 3/27/2009 8:09 AM: > Now I just tried Zenoss core with WMIPerformanceMonitor.zip and I see that it > seems to be able to connect to remote machines using wmi but without a > wmi_proxy. It requires only to specify an user in zProperties tab. > So...the question about Zenoss and wmi are: > > a) How zenoss is able to directly query remote servers using wmi? I mean: > zenoss is linux based so I don't understand how it is possible. Magic... No, they've just extended Samba 4 to speak WMI using Microsoft's protocols on Linux. So no proxy necessary, and it's talking the same language more or less (there have been bugs of course). > b) Is the password sent by Zenoss on the network encrypted? This question is > strictly related to the security problem I just exposed because I'm trying a > product that could satisfy high security levels. As far as I know, it's using MS security standards - it's going to sent the password in the same way that Windows does, so if NTLM or whatever your particular Windows box does is secure enough, then Zenoss will be. The problem is that currently Zenoss stores the password in some events/alerts/e-mails in plain text on the server. There is both configuration to hide that info in events/alerts/e-mails, as well as work to handle this better out of the box. > c) I suppose that with Zenoss would be there the same problem about the > user's rights so...a possible workaround could be use an agent (passive > check) only on the domain controllers (insted wmi query). Is it possible? Zenoss doesn't yet really do agents per se - though you could I suppose write one that talks ssh or whatever you like to get to Zenoss. Simpler is to use Domain Admin, but search the forums, there are instructions (somewhat spread out) on how to use a much more limited domain user account as long as you provide the proper WMI permissions to that account in Active Directory or on the local computers. > > > 2) Recurring Scheduled downtime > One of the missing future of groundwork/nagios is the ability to schedule > recurring downtime. Does Zenoss provide this future? Yes, if I understand correctly. Zenoss has maintenance windows, where it will (if you set up alerts to use that as a filter - they do by default, but you could turn that off) not generate e-mails etc. But I believe events do still get logged in the web UI. > > 3) Scheduled downtime and Availability reports > Creating a "service" or "host" report, Groundwork shows for each main status > (OK,CRITICAL,WARNING,UNKNOWN): > total > scheduled > unscheduled > > This is very important because I need to distinguish between "scheduled and > unscheduled". Does Zenoss works in this way? Not out of the box. I believe you could do this, but I'm not really sure how - someone more versed in custom reports might expand here. > > 4) RRD files management > Groundwork has 2 big limit with rrd files management. > > a) rrd files creation > It create the rrd file in this way: > host_servicename > When you rename an host or a service you lost statistics. Really is possible > to restore them but you must manually change some entries in the database, > rename the files etc.. > This is not so easy, takes a long time and cause a small "hole" in the graph. > Is Zenoss afflicted by this problem or does it handle this problem a little > better? Unfortunately, Zenoss is afflicted by this problem, it uses the host name in the path of the rrd files: $ZENHOME/perf/devicename so you'd likely have to do something similar. Keeping RRD files is a bit of a pain in several instances, you've highlighted one. The other is if you decide you want to change the RRD graphing basics, such as the time to keep data, or the graph definition in any way. > > b) Dynamic Performance graph creation > With Groundwork you can create a template for each service or host but there > is a problem: > supposing yours windows or linux servers have not the same configuration (for > exmaple you have some linux servers with /xxx and others with /yyy) you must > arrange the template. To sum up the performance graph creations is not really > dynamic. > Does Zenoss work in the same way? I'm afraid I'm not really following, but Zenoss lets you make your templates as dynamic as you can? Generally you make a template that matches either a class of devices, or a single device. If your configuration is different, you would need another template. There is some dynamic choosing, but I'm not sure you can extend this in a way that you'd want - see: http://www.zenoss.com/Members/cluther/how-zenoss-chooses-templates/ > > 5) High availability package module > In the "product comparison" I saw that Zenoss Enterprise has an "High > Availability Package" but I have not understood how it works. > Is it an "out of the box solution" and is it simple to configure or does it > require external packages? How does it work? I don't use Enterprise, so I can only speculate, but I believe it would be a Zenpack (that's how they add on the features of Enterprise, the Core is, as implied by the name, the same as Zenoss Core). > > 6) Custom email notification for each service > Groundwork send a specific email for each service but it use a template with > some variables so the email contains just the "Hostname", the "servicename" > and the status (warning,critical etc..). I would like to send some additional > informations in the body for each service. Is possible with Zenoss? Or does > it works in the same way of nagios? You can use variables, and for each alerting rule you create, you can customize your message. You can also customize your events that can get put into the e-mal. There is pretty much a sensible default, but you can go wild customizing it. Note, each Zenoss User account gets to define their own alerting rules (or a Zen Manager can do it for them, but they need to do it for each user I think, or each group). > > 7) Monitor *unix servers with ssh > With groundwork it is quite simple: > the first time is necessary to create ssh key and copy it on remote hosts. > Then you need to configure ssh on remote hosts ("PubkeyAuthentication") and > copy some plugions from nagios/libexec. > How does it work with Zenoss? Zenoss supports SSH monitoring using Zenoss Plugins, or your own set of scripts to run on the remote system and the Zenoss Server. By default, it's a bit limited in what it pulls back. In "Blue Crab", the upcoming 2.4 release (in beta now), SSH monitoring can do anything out of the box that SNMP monitoring does. It works similarly, you can use keys, or I believe you can save the password in the zProperties for the device/device class. I don't do SSH, so someone else may correct me here. > > 8) Tools like Nagvis, webinject and Network Whatermap > Groundwork incluse these wonderful tools. > While you can live without Nagvis and Network Whatermap, in my opinion > WEBINJECT is absolutely necessary in order to test web applications and to > create performance graphs. > Does Zenoss provide these tools? I have no idea what these tools do. Zenoss does provide Zenpack API to extend Zenoss, it also supports Nagios plugins for performance monitoring. > > 9) Performance comparison > Because groundwork is based on nagios (now nagios 3) you can find some > performance tests. Is there a Zenoss comparison? I mean: Nagios 3.0 claims a > 30/50 % increase in performance so...for example with a 3.0 ghz cpu you can > monitor more than... > Is there a Zenoss test like this? Not really, best I've seen in the forum thread on "size of community installs": http://forums.zenoss.com/viewtopic.php?t=6230 Generally Zenoss is rather resource hungry. I believe one collector also tops out at about 1000 devices, though that varies a LOT based on what data points you're collecting (SNMP, WMI, Syslog, Traps, etc) and how fast. ZenSyslog is not as robust as syslog-ng, and isn't going to easily replace that if you want to send ALL syslogs into ZenSyslog. Zenoss is heavily I/O bound, but can use Memory for caching if available, and can easily use a bunch of Memory. CPU use depends on a bunch of factors, but in my experiance, it's less often an issue than I/O speed followed by Memory. You'll probably want RAID or SSD storage. > > 10) Active and passive checks. > Groundwork suggest to NOT use active checks (wmi is an active check) if you > have hundreds of devices but instead to use passive checks (the agent > installed on remote hosts) > What does Zenoss suggest about this? Zenoss doesn't have an agent yet in that way, but yes, so called Active Checks can take time. You sort of have to set up and find out how much load you'd get from Zenoss, and tune from there. There is no great rule of thumb. Certainly, if you can get it from SNMP, or just need uptime so can use ping checks, that's going to be faster than doing WMI or a Nagios plugin or running a script you create to get perf data. > > > > I apologize for this long questions but I think a kind reply would be > appreciated by all Zenoss's newbie like me and people that would like to have > a summary of the main difference between Zenoss and Nagios > > Thanks > You're welcome. -- James Pulver Information Technology Area Supervisor LEPP Computer Group Cornell University _______________________________________________ zenoss-users mailing list [email protected] http://lists.zenoss.org/mailman/listinfo/zenoss-users
