[07-Jan-2010 09:59:49] <mrchippy> yo, everybody. this is Chip. mattray is out of town cuz he's going to the bcs champ game tonight.
[07-Jan-2010 10:00:13] <rmatte> Rolfs: right, so add a command type datasource and then create the datapoints under it
[07-Jan-2010 10:00:28] <rmatte> Rolfs: make sure that they are named exactly after the keys
[07-Jan-2010 10:00:31] <rmatte> case sensitive
[07-Jan-2010 10:00:37] <Rolfs> yeah, that part I understand
[07-Jan-2010 10:00:39] <rmatte> if Key1 is CPU1 then make the datapoint called CPU1
[07-Jan-2010 10:00:44] <Rolfs> but I want to make it more easy
[07-Jan-2010 10:00:52] <smcingvale> Hi Chip, I want to make a multi-graph report for diskspace. I'd prefer to do it by partition. How does this work? If i set it up like my filesystem template (usinig used_bloakc) the multi-graph report shows up blank. Is this even possible?
[07-Jan-2010 10:01:14] <rmatte> Rolfs: If you're talking about having Zenoss dynamically create datapoints based on the number of keys, no, that's not possible
[07-Jan-2010 10:01:32] <Rolfs> i would like to not add the datapoint manualy for every value
[07-Jan-2010 10:01:56] <Rolfs> or key
[07-Jan-2010 10:02:22] <rmatte> well, how many keys do you have to add?
[07-Jan-2010 10:02:32] <rmatte> or datapoints rather
[07-Jan-2010 10:03:19] <willwh> hi guys :)
[07-Jan-2010 10:03:24] <twm1010> Rolfs: The whole point of a template is to only have to do it once...
[07-Jan-2010 10:03:37] <willwh> ipservice time outs.... where might those be? :)
[07-Jan-2010 10:04:04] <Rolfs> Well.. we have a lot of different and special-made munin scripts
[07-Jan-2010 10:04:11] <mrchippy> smcingvale: digesting and looking...
[07-Jan-2010 10:04:20] <smcingvale> cool, thanks.
[07-Jan-2010 10:04:37] <Rolfs> to say it like that. We have a lot of keys... some munin graphs generates a lot of keys dependent on the data.. is there a zendmd way to f.eks add them ?
[07-Jan-2010 10:05:28] <Rolfs> twm: the next problem would be that I cannot add the same template more than once on one server with different parameters.. afik..
[07-Jan-2010 10:05:53] <Rolfs> f.eks 4-5 different databases..
[07-Jan-2010 10:07:10] <Rolfs> how about having a button in zenoss saying.. Try to auto-fetch datapoints.. and then let you edit them after..
[07-Jan-2010 10:07:13] <twm1010> i see your point, some command datasources for monitoring a proprietary database will vary from server to server unless you're monitoring a cluster or mirrored pair
[07-Jan-2010 10:08:36] <Rolfs> I have 110 templates in the /Devices and using them and binding them to several servers/classes
[07-Jan-2010 10:10:32] <Rolfs> I think I found a script for copying templates with zendmd on the zenoss page.. but I havent looked into how to do it per datapoint.. any good suggestions there?
[07-Jan-2010 10:11:05] <Rolfs> <3 zendmd
[07-Jan-2010 10:12:23] <mrchippy> smcingvale: can you give me the steps you're following? which template are you using (the straight-up FileSystem)?
[07-Jan-2010 10:15:18] <smcingvale> I'm currently using the default FileSystem template that uses the HOST-RESOURCES mib. This works fine when I browse to the OS tab for each device. I can see each partition's usage, etc. I want to collect a bunch of devices filesystemm usage, by partition, into a multi-graph report. All I did was create a report using the usedBlocks datapoint (basically copied whats in the filesystem template). I get nothing in the report when I do this...I think th
[07-Jan-2010 10:17:44] <smcingvale> I'd ideally like to have each partition as a separate graph with about 10 devices on the graph. That way I can see which devices are deviating from the norm, etc. Not sure how to do this, thought I could do it like the filesystem template but not sure if there's something more involved or not.
[07-Jan-2010 10:17:47] <twm1010> you did reference the datasource, and the datapoint right?
[07-Jan-2010 10:19:00] <twm1010> I would think in your collections, you'll have to choose the components themselves, not the nodes, i could be wrong
[07-Jan-2010 10:20:01] <mrchippy> smcingvale: i think it's one or the other, unfortunately. a datapoint for each device in a bunch or a datapoint for each component on a device. i'll fool around some more here and see if i can cook something up.
[07-Jan-2010 10:20:43] <mrchippy> smcingvale: (and you're looking for a datapoint for a named component on each device in a group of devices, right?)
[07-Jan-2010 10:21:18] <smcingvale> mrchippy: correct.
[07-Jan-2010 10:22:10] <smcingvale> twm1010: Is there a way to referencec the components themselves (ie not the nodes) in a multi-graph report?
[07-Jan-2010 10:23:02] <twm1010> I would think so, just like you can reference interfaces specifically
[07-Jan-2010 10:24:13] <smcingvale> How do I set this up in the report? Is there a way to make a collection that references components (or interfaces)?
[07-Jan-2010 10:25:05] <smcingvale> twm1010: Think I just figured it out...didn't notice the Item Type of Device/Component...will give this a shot.
[07-Jan-2010 10:25:09] <twm1010> when you define the collection
[07-Jan-2010 10:25:14] <twm1010> there is n option for componenet
[07-Jan-2010 10:25:19] <twm1010> you can choose a disk
[07-Jan-2010 10:25:37] <twm1010> When I was creating a report mixing bandwidth and apache/iis stuff, i had to reference the interfaces directly
[07-Jan-2010 10:25:43] <twm1010> in a separate collection
[07-Jan-2010 10:25:59] <twm1010> Its pretty granular, but not as automatic as we'd like, but it gets it done.
[07-Jan-2010 10:26:06] <mrchippy> yeah, you can manually build the list.
[07-Jan-2010 10:26:26] <smcingvale> Awesome, thanks guys. I'll play around with this for a bit.
[07-Jan-2010 10:26:58] <fr500> hi
[07-Jan-2010 10:27:08] <fr500> i'm testing zenoss
[07-Jan-2010 10:27:20] <fr500> I added two devices on two locations with the addresses set
[07-Jan-2010 10:27:29] <fr500> they appear on the map fine
[07-Jan-2010 10:27:35] <fr500> but there is no line connecting them
[07-Jan-2010 10:27:50] <fr500> I already set the property to draw links to true
[07-Jan-2010 10:28:02] <cparlette> There are four conditions that must be met for a line to be drawn.
[07-Jan-2010 10:28:02] <cparlette> 1. DeviceX and DeviceY both have network interfaces on a shared IP subnet.
[07-Jan-2010 10:28:02] <cparlette> 2. The shared IP subnet has its zDrawMapLinks property set to true.
[07-Jan-2010 10:28:02] <cparlette> 3. DeviceX and DeviceY are in LocationA and LocationB respectively.
[07-Jan-2010 10:28:02] <cparlette> 4. LocationA and LocationB are shown on the current Google Map.
[07-Jan-2010 10:29:16] <Rolfs> fr500: have you put up the router in your zenoss?
[07-Jan-2010 10:29:21] <fr500> Location A and Location B inside a common organizer
[07-Jan-2010 10:29:37] <smcingvale> mychippy,twm1010: That worked beautifully. Thanks.
[07-Jan-2010 10:29:46] <fr500> Rolfs, yes, the devices are correctly modeled and set to the corresponding location
[07-Jan-2010 10:31:29] <smcingvale> My other question is: Is there anyway to import a bunch of OIDs into a template...either through the interfacce or programatically. I have a mib file containing some custom oids for custom apps...want to import about 200 of these into a template so we can use them to monitor. Don't want to have to go one by one. I've already imported the mibs...just need to use them to actively monitor...
[07-Jan-2010 10:32:16] <twm1010> yikes.... 200 oids?!?
[07-Jan-2010 10:32:33] <twm1010> i think you'd have to script it
[07-Jan-2010 10:33:12] <smcingvale> That's what I was thinking. Don't have a lot of experience and couldn't find a lot in the docs. Could you point me in the right direction on how to get started writing a script...or, are there any currently out there?
[07-Jan-2010 10:33:27] <Rolfs> Oh.. that would be almost like the same question as I have.. If you program it.. I would like to know how it's done
[07-Jan-2010 10:34:13] <smcingvale> If I get it working I'll definitely post it in the forums...just need to know where to start :). This is basically what zenpacks do...
[07-Jan-2010 10:35:08] <Rolfs> probably have to use zendmd, patch it so you have a tab compleation is my first tip
[07-Jan-2010 10:35:11] <twm1010> i dunno if you can design a template from zenDMD
[07-Jan-2010 10:35:14] <fr500> Rolfs, well one of the devices is down, maybe that's why the line isn't drawn?
[07-Jan-2010 10:35:28] <twm1010> but, it might be possible, but complex
[07-Jan-2010 10:35:34] <twm1010> not every datapoint is the same type
[07-Jan-2010 10:35:56] <twm1010> I think it would depend on just how many times you're going to have to do it.
[07-Jan-2010 10:36:10] <twm1010> if you only need one template with those oid's, I'd suck it up and start typing.
[07-Jan-2010 10:36:23] <twm1010> but in Rolfs case where each template is different, yikes
[07-Jan-2010 10:36:48] <twm1010> a slick windows user using AutoIT might pull it off by manipulating the gui, but zenoss isn't always so snappy :)
[07-Jan-2010 10:37:15] <twm1010> so, zenDMD it is, but i don't know if its even capable of creating a template,
[07-Jan-2010 10:37:35] <smcingvale> What about checking out some zenpack source? Don't a lot of the zenpacks simply create device template and stick a bunch of oids in there...I just want that capability with any generic oid list...
[07-Jan-2010 10:37:40] <Rolfs> mm I would just like to be able to have a button next to the 'test' button saying. Try to add datapoints from output of command
[07-Jan-2010 10:37:57] <smcingvale> OK, I can create the template myself...then use the script to import the oids. Think that would work?
[07-Jan-2010 10:38:15] <twm1010> Rolfs: thats a good suggestion, you should open up a trac ticket for a feature enhancement
[07-Jan-2010 10:38:54] <twm1010> especially for database monitoring on assimilar targets, it could be really useful
[07-Jan-2010 10:40:35] <Rolfs> seems to me that if you want to create a zenpack out to import the OIDs, you will have to create a objects.xml file .. which I have never tried to do..
[07-Jan-2010 10:44:43] <mrchippy> you could do either. you can create templates/ds/dp's in zendmd and export them for versioning in the zenpack (objects.xml), or you could create objects.xml directly. Either will work. But I definitely agree that a visual way to do this easily would be a kicka$$ feature.
[07-Jan-2010 10:44:44] <Rolfs> I think I whould have tried to copy a small ZenPack containing a small amount of OIDs and modified it and then imported it.. but after a backup of the zenoss..
[07-Jan-2010 10:44:48] <rmatte> eugh wtf... I'm getting utilization alerts for interfaces even though I have interface utilization thresholds disabled
[07-Jan-2010 10:45:36] <smcingvale> I'll give that a shot and see what I can come up with.
[07-Jan-2010 10:45:37] <mrchippy> rmatte: which daemon? snmp or command?
[07-Jan-2010 10:45:56] <rmatte> snmp
[07-Jan-2010 10:46:10] <rmatte> it just started happening
[07-Jan-2010 10:46:39] <mrchippy> were the thresholds just disabled recently?
[07-Jan-2010 10:46:46] <rmatte> nope
[07-Jan-2010 10:46:50] <rmatte> it's been over a month
[07-Jan-2010 10:47:09] <rmatte> and it's only happening on one specific device for one specific port
[07-Jan-2010 10:47:46] <Rolfs> ImportError: No module named ZenCollector.daemon - Is that a new module in zenoss 2.5 ?
[07-Jan-2010 10:47:56] <mrchippy> huh. can you run a "zenperfsnmp -d <deviceid> -v10" and post the log to the forums. i'll take a look
[07-Jan-2010 10:48:14] <mrchippy> Rolfs: yes it is
[07-Jan-2010 10:48:18] <mrchippy> where'd you see that?
[07-Jan-2010 10:48:20] <rmatte> mrchippy: k
[07-Jan-2010 10:48:37] <Rolfs> mrchippy: is it a way to fix it when using a plugin in a 2.4.1 version?
[07-Jan-2010 10:50:25] <rmatte> hmmm, there's really nothing in the zenperfsnmp run that would explain it
[07-Jan-2010 10:50:32] <rmatte> I don't even see the threshold being triggered
[07-Jan-2010 10:52:03] <mrchippy> rmatte: well, that stinx. hmmm...
[07-Jan-2010 10:52:22] <mrchippy> Rolfs: which plugin? where'd you get it?
[07-Jan-2010 10:52:58] <Rolfs> I get it when I restart zenoss. I have added the ZenPacks.community.ZenODBC-2.1.1-py2.4.egg plugin in my zenoss 2.4.1
[07-Jan-2010 10:53:16] <Rolfs> File "/opt/zenoss/ZenPacks/ZenPacks.community.ZenODBC-2.1.1-py2.4.egg/ZenPacks/community/ZenODBC/zenperfodbc.py", line 36, in ?
[07-Jan-2010 10:53:16] <Rolfs> from Products.ZenCollector.daemon import CollectorDaemon
[07-Jan-2010 10:53:24] <mrchippy> Rolfs: ok, i'll take a look at that bad boy
[07-Jan-2010 10:54:32] <rmatte> mrchippy: oh well, maybe I just need to restart Zenoss on that server, it was crazy slow earlier since we were gathering data for automated reports from it, plus it was getting hammered with monitoring
[07-Jan-2010 10:55:09] <Rolfs> is it a way to add the CollectorDaemon somehow, or would that mess up a lot?
[07-Jan-2010 10:55:41] <willwh> hey guys, sorry to ask this again, but, I am trying to find the timeout settings for ipservice/http
[07-Jan-2010 10:55:47] <mrchippy> rmatte: maybe the stress caused something to lose the configuration update? but just for one device?
[07-Jan-2010 10:56:08] <willwh> one of my boxes reported http being down; event was cleared 1 minute late.... I just wanted to tune it a little
[07-Jan-2010 10:56:10] <rmatte> mrchippy: maybe
[07-Jan-2010 10:56:58] <rmatte> seems to have settled down, I'll wait and see if it happens again
[07-Jan-2010 10:57:14] <mrchippy> Rolfs: no, the CollectorDaemon stuff comes from an overhaul of the daemons. we're moving all the daemons over to this framework. you'd wind up having to do an upgrade.
[07-Jan-2010 10:58:01] <Rolfs> willwh: try to look under the OS tab.. The Event Might tell you more about how the event was discovered.
[07-Jan-2010 10:58:24] <Rolfs> willwh: You could turn off monitoring of the http service under OS-tab
[07-Jan-2010 10:58:25] <mrchippy> Rolfs: the problem is that apparently the setup.py allowed that zenpack to be installed when it's not compatible. the version compatibility string probably didn't get updated :(
[07-Jan-2010 10:59:25] <Rolfs> mrchippy: I guess so..
[07-Jan-2010 10:59:46] <mrchippy> Rolfs: yikes, the COMPAT_ZENOSS_VERS is ">=2.5". even worse, maybe zenoss's compatibility check is not working :O
[07-Jan-2010 10:59:49] <willwh> Rolfs: great thank you
[07-Jan-2010 11:01:22] <theacolyte-> Finding SNMP to be even harder to work with than WMI
[07-Jan-2010 11:01:55] <eidolon> hey folks, what's the seperator for Nagios format checkers for multiple data sources? I have OK|Field=somevalue
[07-Jan-2010 11:01:59] <Rolfs> theacolyte-: why is that?
[07-Jan-2010 11:01:59] <eidolon> wht if ther'e more than one Field?
[07-Jan-2010 11:02:23] <willwh> snmp gives me everything I need
[07-Jan-2010 11:02:26] <Rolfs> asdasdasd |key=value key2=value2
[07-Jan-2010 11:02:27] <theacolyte-> Rolfs: I have a template that is working just fine in Cacti that I tried to move over - same OID's and the numbers are way off
[07-Jan-2010 11:02:32] <eidolon> space. okay, thanks
[07-Jan-2010 11:03:49] <theacolyte-> Unfortunatley I deleted the old template of of cacti to try a different one, and this one doesn't even translate over - they do some weird OID manipulation I haven't figured out yet
[07-Jan-2010 11:03:52] <rmatte> theacolyte-: you sure cacti isn't doing some sort of processing on the numbers?
[07-Jan-2010 11:04:05] <rmatte> theacolyte-: or maybe you need to change the format options for the graph in Zenoss
[07-Jan-2010 11:04:32] <theacolyte-> rmatte: I'm going to figure out how to get these graphs back together so I can troubleshoot and I'll ping ya
[07-Jan-2010 11:04:45] <rmatte> theacolyte-: it's technically impossible for the values to be noticeably different if you're using the exact same OID as cacti is
[07-Jan-2010 11:04:54] <rmatte> theacolyte-: have you tried snmpwalking the OIDs to see what you get?
[07-Jan-2010 11:04:57] <theacolyte-> That's what I'm saying.
[07-Jan-2010 11:05:09] <theacolyte-> I did
[07-Jan-2010 11:05:14] <rmatte> ...and?
[07-Jan-2010 11:05:15] <theacolyte-> Same values it Zenoss was reporting
[07-Jan-2010 11:05:24] <theacolyte-> s/it/that/
[07-Jan-2010 11:05:30] <rmatte> yeh, then Cacti is doing something else to the values obviously
[07-Jan-2010 11:05:35] <rmatte> if they aren't showing up as you'd expect
[07-Jan-2010 11:05:57] <theacolyte-> I don't even know how to figure it out, but at the moment I don't have it in place to troubleshoot
[07-Jan-2010 11:06:21] <rmatte> you'd need to know what it's doing to the values, once you know that it'd be simple from then on
[07-Jan-2010 11:07:35] <theacolyte-> yeah
[07-Jan-2010 11:07:54] <theacolyte-> I tried doing some RPN, but then I ran into an issue where the value actually wasn't changing
[07-Jan-2010 11:08:13] <theacolyte-> But, this new cacti template I'm trying to adapt to zenoss is weird and I can't figure out how they have the OIDs set so I'm walking
[07-Jan-2010 11:08:25] <rmatte> lol
[07-Jan-2010 11:08:49] <theacolyte-> The problem with equallogic SAN's is the OIDs change on each chassis (or, maybe, model)
[07-Jan-2010 11:09:12] <willwh> Rolfs: had a really good hunt around - I still don't see *why* this report http as being down - it doesn't explicitly say timeout... but I think so
[07-Jan-2010 11:09:45] <willwh> nothing in the event log on the box, so, I'd like to tune the timeout
[07-Jan-2010 11:09:49] <rmatte> willwh: telnet to port 80 on whatever device and type a bunch of garbage then hit enter and make sure you actually get some sort of return
[07-Jan-2010 11:09:55] <Rolfs> willwh: did you move it to history?
[07-Jan-2010 11:10:01] <rmatte> willwh: Zenoss does more than just checking to see if port 80 is open
[07-Jan-2010 11:10:07] <willwh> ah
[07-Jan-2010 11:10:25] <willwh> well it's not a web service that runs there.... but that is the 1 and only report of it being down
[07-Jan-2010 11:10:49] <willwh> we service streams on port 80 from most of our boxes, so, the latency to our boxes can be quite high
[07-Jan-2010 11:11:01] <Rolfs> them move the event to history if you have removed it from beeing monitored in the os-page
[07-Jan-2010 11:11:01] <rmatte> I see
[07-Jan-2010 11:11:09] <willwh> I've already tuned the ping timeout to 5seconds (yes I need that - at peak some boxes take aorund 4.5s to respond, although they are working quite happily)
[07-Jan-2010 11:12:00] <willwh> just serving video at 800mbps
[07-Jan-2010 11:12:05] <willwh> :)
[07-Jan-2010 11:12:43] <rmatte> Zenoss connects to port 80 and expects some sort of http return
[07-Jan-2010 11:12:56] <Rolfs> and you have debugged it with trying to fetch the data with the zencommand run -d device or somemthing like that?
[07-Jan-2010 11:12:57] <rmatte> so if it's not seeing that then it's going to think it's down
[07-Jan-2010 11:12:59] <willwh> which it will recieve
[07-Jan-2010 11:13:08] <willwh> (this is a windows media services box)
[07-Jan-2010 11:13:13] <willwh> so it does give an http response
[07-Jan-2010 11:13:17] <rmatte> k
[07-Jan-2010 11:13:28] <mrchippy> smcingvale: check out thread/12475
[07-Jan-2010 11:13:29] <willwh> like I said - I've been monitoring this box for ~3days
[07-Jan-2010 11:13:31] <willwh> this is the only event
[07-Jan-2010 11:13:43] <willwh> it was a "hiccup" - looks like a timeout to me
[07-Jan-2010 11:13:49] <Apachez> are there any template or such for selfmonitoring on the zenoss ?
[07-Jan-2010 11:13:55] <Rolfs> could it be that all the daemons where busy serving.. and none where available to answer your zenoss check?
[07-Jan-2010 11:14:04] <Apachez> or do I have to install snmpd manually on the zenoss server first?
[07-Jan-2010 11:14:16] <willwh> Rolfs: nope
[07-Jan-2010 11:14:40] <willwh> 40mbps out / only serving aroun 1800 audio streams right now
[07-Jan-2010 11:14:50] <willwh> is there a timeout I can configure on those checks?
[07-Jan-2010 11:15:26] <theacolyte-> wow
[07-Jan-2010 11:15:33] <theacolyte-> someone that actually uses windows media services
[07-Jan-2010 11:15:38] <theacolyte-> which version?
[07-Jan-2010 11:15:39] <willwh> unfortunately
[07-Jan-2010 11:15:43] <willwh> we use a bunch of stuff
[07-Jan-2010 11:16:08] <theacolyte-> I ran it for awhile here, it wasn't tooooooo bad to deal with, but it felt like an awful, awful hack
[07-Jan-2010 11:16:19] <willwh> yeah
[07-Jan-2010 11:16:26] <willwh> using Wowza Media server to dist most stuff
[07-Jan-2010 11:16:32] <willwh> and icecast
[07-Jan-2010 11:16:36] <willwh> and darwin
[07-Jan-2010 11:16:43] <theacolyte-> What, if you don't mind, does your company use it for?
[07-Jan-2010 11:16:43] <QubeZ> man this asterisk monitoring zenpack is just killing my server :(
[07-Jan-2010 11:16:44] <willwh> some shoutcast setups too actually
[07-Jan-2010 11:16:59] <willwh> theacolyte-: we are a streaming media host :)
[07-Jan-2010 11:17:16] <theacolyte-> willwh: Powerstream? :P
[07-Jan-2010 11:17:22] <willwh> no :)
[07-Jan-2010 11:17:27] <willwh> we are a small outfit ;)
[07-Jan-2010 11:17:30] <theacolyte-> ah
[07-Jan-2010 11:17:44] <theacolyte-> I had looked into outsourcing it, but ended up doing it internally
[07-Jan-2010 11:17:49] <theacolyte-> powestream was a company I looked into
[07-Jan-2010 11:18:38] <Rolfs> QubeZ: why is it killing your server?
[07-Jan-2010 11:18:57] <theacolyte-> This could all be solved if Zenoss had a button: Import from Cacti XML template
[07-Jan-2010 11:20:28] <QubeZ> Rolfs sometimes python doing an AMI check of the asterisk statistics just hangs and CPU goes to 100%
[07-Jan-2010 11:20:42] <QubeZ> i have 6 asterisk servers monitored, just two python scripts running again two of my servers was at 100% cpu
[07-Jan-2010 11:20:46] <QubeZ> i killed the process
[07-Jan-2010 11:21:26] <theacolyte-> Just ask your users this:
[07-Jan-2010 11:21:50] <theacolyte-> "Look guys, would you rather we monitor for it to be up and down, and keep it down consistantly, or have it down ocassionally and not know?"
[07-Jan-2010 11:23:54] <rmatte> QubeZ: thought there was a brand new Asterisk ZenPack released recently that fixed some of those performance issues?
[07-Jan-2010 11:25:37] <Rolfs> i find it quite usefull to have the "zencommand run -v10 -d ${device/id}" as a command to run on the server from the webgui
[07-Jan-2010 11:26:13] <QubeZ> rmatte the ssh one? its limited in what it reports i believe
[07-Jan-2010 11:35:19] <Rolfs> was it suddently quiet here?