[2009-06-11 10::01:43] ** You changed the topic to "Zenoss Developers are here".
[2009-06-11 10::03:02] cgibbons: *golfclap*
[2009-06-11 10::03:10] mrayzenoss: good morning/afternoon/evening everyone, cgibbons and I will be here to answer your questions
[2009-06-11 10::03:45] rmatte: good morning Matt
[2009-06-11 10::03:50] cgibbons: bwest: yeah, rocket's right. it's much better to use the built-in WMI support to do your own modeling plugins if you need.
[2009-06-11 10::04:51] bwest: I am using 2.4.1
[2009-06-11 10::06:39] cgibbons: not a good way to do it, really. winexe isn't a good tool. zenoss ships with wmic as a command-line tool that you can use, plus you can run WMI queries directly from the python code.
[2009-06-11 10::06:59] ** jplouis joined the chat room.
[2009-06-11 10::08:15] bwest: So how do I get to the command line when modeling to use wmic
[2009-06-11 10::08:21] rmatte: ok, so my main question for the developers: On the OS tab, port status appears to be determined during modelling. Why is no polling done on monitored interfaces each polling cycle to check status? I'm having to use a threshold to properly monitor administrative status on ports, and it doesn't update the status on the OS tab. Ports stay showing as up when they should be showing as down, and vice versa.
[2009-06-11 10::08:45] cgibbons: for example I just did this here: wmic -U 'zenoss\administrator' //hostname 'select SerialNumber from Win32_SystemEnclosure'
[2009-06-11 10::10:10] ** magnachef_ left the chat room.
[2009-06-11 10::10:13] mrayzenoss: cgibbons: this one? http://www.zenoss.com/community/projects/zenpacks/windowswmidevicetemplate
[2009-06-11 10::10:25] cgibbons: look at WMIPlugin.py in Products.ZenWin for somethinng you can extend in your own modeling plugins to do WMI queries
[2009-06-11 10::10:41] bwest: ok thx
[2009-06-11 10::11:58] cgibbons: the way WMIPlugin.py is done you don't have to worry about executing the query yourself, you just override the queries() method with the WMI queries you want it to execute
[2009-06-11 10::13:07] rmatte: cgibbons: ok, is it something that's being looked at for the future at all?
[2009-06-11 10::14:07] cgibbons: it hasn't been escalated to the point where the enterprise customers have asked for it (surprisingly). i'm surprised it doesn't come up more.
[2009-06-11 10::14:31] ckrough: Not very manageable on a large scale.
[2009-06-11 10::17:01] ** sergeymasushko left the chat room.
[2009-06-11 10::18:07] bwest: If I have an environment that has about 350 components it's monitoring, running 2.4.1 on linux what server requirments shold be large enough? The components are 200 servers and 150 switches and routers
[2009-06-11 10::18:24] rmatte: I'm just looking at this from the perspective of Zenoss' campaign to win over people from the other larger monitoring companies/tools, but if you're lacking basic functionality like this, it doesn't really make a good showing.
[2009-06-11 10::19:48] cgibbons: there's a chart we have on teh website with requirements... let me find it
[2009-06-11 10::20:19] ckrough: bwest: depents on the number of things you are collecting, how long you keep the rrd data, and how frequently you collect it
[2009-06-11 10::21:01] bwest: they are interested in baselining and they are monitoring this data around the clock
[2009-06-11 10::22:56] mrayzenoss: make sure you're on a fast filesystem
[2009-06-11 10::23:01] cgibbons: mattray, do you have any stats available with some of our really big customers?
[2009-06-11 10::23:15] mrayzenoss: well, the largest community install 1 box is 8500 devices
[2009-06-11 10::23:26] ckrough: 1 box?
[2009-06-11 10::23:36] mrayzenoss: yeah, 8-way 32 gigs of RAM
[2009-06-11 10::23:54] rocket: ckrough: devices isnt really indicitive of the amount of data they are collecting
[2009-06-11 10::23:54] mrayzenoss: someone else is doing 10K
[2009-06-11 10::24:02] ckrough: I can only get 1600 devices on that spec, with 6 15k RAID10 storage config
[2009-06-11 10::24:08] rocket: ckrough: need to look at the # of datapoints collected
[2009-06-11 10::24:19] ckrough: 75k OIDs per box on that spec
[2009-06-11 10::25:01] cgibbons: I suspect a lot of the large upcoming installations will be doing mostly Windows devices using WMI/PerfMon, so that'll change the game some
[2009-06-11 10::25:32] mrayzenoss: if you go to Collectors->localhost->Performance you can see your stats
[2009-06-11 10::27:05] bwest: one collector with 10k
[2009-06-11 10::27:19] rmatte: one collector with 10k is insane
[2009-06-11 10::27:27] ckrough: 10k what
[2009-06-11 10::27:34] Jontler: devices
[2009-06-11 10::27:37] rmatte: I assumed he meant devices
[2009-06-11 10::27:41] mrayzenoss: yeah
[2009-06-11 10::27:52] rmatte: what kind of devices though?
[2009-06-11 10::27:59] mrayzenoss: cable modems?
[2009-06-11 10::28:14] bwest: no not devices data points
[2009-06-11 10::28:14] rmatte: oh, that's not really incredibly impressive then
[2009-06-11 10::28:24] mrayzenoss: right
[2009-06-11 10::28:44] ckrough: disk IO seems to be the killer. anyone runnign this on SSD yet?
[2009-06-11 10::29:41] cgibbons: it scales nicely horizontally for that issue, since data (sans events) is kept on each collector.
[2009-06-11 10::29:45] rmatte: I'm polling like 18.7k datapoints on my largest install so far, but I'm going to have much larger projects coming down the pipe soon
[2009-06-11 10::30:27] bwest: but for 10K datapoints I should just look at the sizing chart?
[2009-06-11 10::30:53] mrayzenoss: here's the bragging thread for Core installs: http://forums.zenoss.com/viewtopic.php?t=6230
[2009-06-11 10::31:24] ckrough: >430k OID on my biggest install
[2009-06-11 10::31:45] ** mistich joined the chat room.
[2009-06-11 10::31:51] ** willwh joined the chat room.
[2009-06-11 10::31:53] mrayzenoss: heh, I'm doing about 110 on my home install
[2009-06-11 10::33:00] cgibbons: hrmph only 220 on my home setup, which i haven't looked @ in months
[2009-06-11 10::33:25] ckrough: Yeah, I plan to set it up at home to watch my total usage with NetFlix... b/w caps are coming...
[2009-06-11 10::33:30] cgibbons: doh it's only 2.3.1, bad me
[2009-06-11 10::33:48] rocket: cgibbons: heh I dont have as many datapoints .. but a fairly large number of zenpacks installed
[2009-06-11 10::33:59] mrayzenoss: my home install is where the source build on PPC gets tested
[2009-06-11 10::34:05] ** chudler joined the chat room.
[2009-06-11 10::34:11] cgibbons: you still have that powerbook??
[2009-06-11 10::34:18] mistich: does anyone know the status of IPV6 in zenoss
[2009-06-11 10::34:35] mrayzenoss: mistich: it hasn't gotten prioritized very high
[2009-06-11 10::34:44] ckrough: plays sad trombone
[2009-06-11 10::34:46] mrayzenoss: mistich: there is a HOWTO somewhere for doing IPV6 ping
[2009-06-11 10::35:24] bwest: I am trying to model SIP adapters. SNMP is not available on them, but they can be identified with Mac addresses, can zenoss get this information
[2009-06-11 10::35:25] mistich: yes but what about snmp ssh etc
[2009-06-11 10::35:30] mrayzenoss: I did get some emails from a university in China that said their class project was to port Zenoss to IPV6
[2009-06-11 10::35:54] ckrough: Is there much demand for making reporting more robust? Maybe moving graph definitions to re-usable objects, etc
[2009-06-11 10::36:01] mistich: yes I saw that too has anyone followed up with them
[2009-06-11 10::36:06] mrayzenoss: I did
[2009-06-11 10::36:36] ** sergeymasushko joined the chat room.
[2009-06-11 10::36:38] cgibbons: there's some good porting enhancements coming in the next release actually
[2009-06-11 10::36:43] mrayzenoss: ckrough: there are already some changes going into King Crab for reporting
[2009-06-11 10::36:50] rmatte: mrayzenoss: probably, I'm sure they found out that it was much more complicated than they initially thought
[2009-06-11 10::36:57] mrayzenoss: rmatte: definitely
[2009-06-11 10::37:05] ckrough: mrayzenoss: I will send you swag if you do
[2009-06-11 10::37:09] mrayzenoss: heh
[2009-06-11 10::37:36] rmatte: you'd almost have to have some sort of layer between Zenoss and the device that takes a dummy ipv4 address and forwards the traffic to an ipv6 address and vice versa
[2009-06-11 10::37:50] mrayzenoss: yeah, it's non-trivial
[2009-06-11 10::37:57] mistich: next question moving away from rrds to a relational database
[2009-06-11 10::38:03] ckrough: ^^ this
[2009-06-11 10::38:11] sergeymasushko: can someone help me with the graph issue. I believe it should be simple, but I stack on it. http://forums.zenoss.com/viewtopic.php?t=10008
[2009-06-11 10::38:32] rmatte: Matt, any progress on figuring out what the cause of that zenpack load time issue could be?
[2009-06-11 10::39:07] mrayzenoss: rmatte: jplouis was looking at it, I believe nothing obvious was showing up yet
[2009-06-11 10::39:35] cgibbons: meaning you want to export your raw RRD data into a relational database for custom reporting, mistich, or you want to change the data collection to store in a database directly?
[2009-06-11 10::39:36] rmatte: cool
[2009-06-11 10::39:44] mrayzenoss: mistich: the idea of moving off of RRD does come up, just not sure what we'd move to
[2009-06-11 10::40:12] rmatte: one of our guys here came up with a way to dump an RRD to XML and then pull from there
[2009-06-11 10::40:16] mistich: well for reporting relational database would be nice
[2009-06-11 10::40:35] bwest: I am sorry i am so full of questions, but is it possible to send mysql queries from zenoss to read the amount of data in a table and create events off this?
[2009-06-11 10::40:36] mistich: but want realtime data
[2009-06-11 10::40:37] jplouis: zenpack slowness has to do with the way the relationships are loaded. The zenpack has a relationship to each packable
[2009-06-11 10::40:43] mrayzenoss: bwest: yes
[2009-06-11 10::41:10] cgibbons: I can tell you from experience that dumping that much raw data into a relational database in real-time is not a fun path.
[2009-06-11 10::41:54] mistich: then once in a relational database reporting becomes very easily can use any package to do reporting
[2009-06-11 10::41:54] mrayzenoss: bwest: have you tried the Core MySQL ZenPack?
[2009-06-11 10::42:03] rmatte: bwest, you'd have to write a script for that and use it as a command based datapoint
[2009-06-11 10::42:17] bwest: perfect
[2009-06-11 10::42:18] mrayzenoss: bwest: I just got another massive set of ZenPacks for doing databases yesterday (MySQL, PostgreSQL, MS SQL, ZeoDB)
[2009-06-11 10::42:24] rmatte: although that MySQL ZenPack might be able to do it, haven't looked at it
[2009-06-11 10::42:37] mistich: oh I understand my install is huge but everyday getting asked for more and more reports
[2009-06-11 10::42:40] mrayzenoss: it provides some stats, it may be good enough
[2009-06-11 10::43:02] rmatte: MS SQL you say?
[2009-06-11 10::43:07] mrayzenoss: rmatte: yeah
[2009-06-11 10::43:18] rmatte: I've been waiting for a good community Zenpack for that
[2009-06-11 10::43:28] mrayzenoss: don't like the existing one?
[2009-06-11 10::43:41] bwest: when will that one be published My SQL
[2009-06-11 10::43:57] mrayzenoss: probably over the weekend
[2009-06-11 10::44:22] bwest: excellent, your contest is working
[2009-06-11 10::44:30] ** CamargoBP joined the chat room.
[2009-06-11 10::44:32] mrayzenoss: yeah, we already have the same number of entries as last year
[2009-06-11 10::44:53] mistich: yeah we have about 6 zenpacks one day well publish
[2009-06-11 10::45:07] chudler: I do have at least 4 more to send you, but I have to get time to polish them
[2009-06-11 10::45:10] mistich: well I'll wait till the prizes are bumped
[2009-06-11 10::45:53] mrayzenoss: the current backlog is: PredictiveThreshold, AIX SNMP, Netware, LinuxMonitorAddon, Gentoo, SuSE, Jabber XMMP, RDBMS, MySQL, MS SQL Server, PostgreSQL, ZeoDB
[2009-06-11 10::46:27] mistich: PredictiveThreshold is this holt winters
[2009-06-11 10::46:32] mrayzenoss: yes
[2009-06-11 10::46:40] mistich: already have that one
[2009-06-11 10::46:47] mrayzenoss: I'm comparing the Enterprise one with the Community one now
[2009-06-11 10::47:25] bwest: Are there any good zenpacks to monitor VOIP in core or enterprise?
[2009-06-11 10::47:27] etank: mrayzenoss: i think that the biggest issue for me was getting the Perfmon ZenPack to work
[2009-06-11 10::48:00] mrayzenoss: bwest: the Asterisk one seems to have some issues, but you could probably adapt any Nagios plugin that already exists
[2009-06-11 10::48:05] cgibbons: i may be biased, but if it were me, if a zenpack uses winexe I'd ditch it and never look bck.
[2009-06-11 10::48:06] sergeymasushko: bwest: what do you want to monitor in the VOIP?
[2009-06-11 10::48:42] bwest: Right now they use Cricket to monitor calls per minute users and the such
[2009-06-11 10::49:01] etank: mrayzenoss: is there a PredictiveThreshold community Pack? or just for Enterprise?
[2009-06-11 10::49:13] mrayzenoss: this will be for Core
[2009-06-11 10::49:35] etank: really? how different would it be from the Enterprise one?
[2009-06-11 10::49:50] mistich: what is the difference between them?
[2009-06-11 10::49:53] mrayzenoss: I just set it up yesterday, they take a few days to generate data
[2009-06-11 10::49:58] ** jsynack joined the chat room.
[2009-06-11 10::50:07] mrayzenoss: I've got both of them installed and testing
[2009-06-11 10::50:17] sergeymasushko: bwest: what is the voip platform (sip server ?)
[2009-06-11 10::50:45] bwest: yes
[2009-06-11 10::50:53] rmatte: I have another question... when executing a command based script where you need to pass a password value to it, if the command fails the password gets spit out in plaintext in the command failed event. I logged a ticket for this and was told to "See the ZenPack creator for a fix". I've already tried fixing this ZenPack by using a python script to pull the username and password values from the Zope database, but after extensive testing I have determined that
[2009-06-11 10::50:59] mistich: one thing to look for is low data in the kb holtwinters doesn't like it and sends false alerts
[2009-06-11 10::51:15] rmatte: The only way to possibly fix this security issue is to have Zenoss intelligently sensor password values in command failed events.
[2009-06-11 10::52:04] mrayzenoss: well, I believe the ZenPack in question was one we don't even have access to
[2009-06-11 10::52:25] cgibbons: the reason that was given as the answer rmatte is because there are two ways to deal with it: 1) not use COMMAND datatypes (which is what we mostly do), 2) a command script that needs a password gets it sent via stdin instead of on the command-line
[2009-06-11 10::52:25] mrayzenoss: it's one of James Dastrup's, which we haven't looked at since he hasn't sent it to us for hosting
[2009-06-11 10::52:46] sergeymasushko: bwest: for monitoring sip servers I use a shell command that run SIPP util with the different scenarios... then parse result and make an event if its needed. also I'm monitoring active calls on my cisco AS5300 GWs...
[2009-06-11 10::52:49] rocket: mistich: I wrote the predictive zenpack .. and tried to make it as configurable as possible
[2009-06-11 10::53:12] rmatte: yeh, but it's not specific to that zenpack is my point, this issue pops up with any type of command based poll that needs a password passed to it from Zenoss
[2009-06-11 10::53:21] mistich: rocket: nice
[2009-06-11 10::53:24] rmatte: whether it be a custom built script, or a core Zenpack
[2009-06-11 10::53:29] bwest: serg: sounds interesting, is there a way for you to share more with me
[2009-06-11 10::53:29] cgibbons: there's a mechanism to do it that doesn't require it being sent on the command-line
[2009-06-11 10::54:07] rmatte: which would be using a python script to pull the values directly from Zope, correct?
[2009-06-11 10::54:10] mistich: like that it retofits the rrd but does it also remove the holtwinters if you want to remove it
[2009-06-11 10::54:11] rocket: mistich: so if you are getting false alerts my threshold will also allow you to go and reapply the beta and gamma and window etc
[2009-06-11 10::54:12] sergeymasushko: bwest: a bit busy for now... lets talk about a bit later...
[2009-06-11 10::54:25] bwest: ok
[2009-06-11 10::54:25] rocket: mistich: it does not currently remove
[2009-06-11 10::54:31] mistich: can't wait to look at it
[2009-06-11 10::54:54] rocket: mistich: as I cant think of a good way to have zenoss do that on removal ..
[2009-06-11 10::55:22] rmatte: I've already done that and modified the scripts in 3 different WMI zenpacks to use it, but we had a Zenoss server crash because the retrieval of the password and username values for each poll was way too intensive, if you can suggest a more efficient way of doing it then I'd be happy to modify the Zenpacks in that fashion.
[2009-06-11 10::55:46] mistich: rocket: have run into that before thats why I bring it up
[2009-06-11 10::56:05] chudler: when I need a password, I put it in a file, and read it from there. Not sure how/if that helps you, rmatte
[2009-06-11 10::56:14] rmatte: It's either a little bit of code to sensor the password, or large loss of performance in the scripts
[2009-06-11 10::56:17] rocket: mistich: it might be something that I could make a command line to do
[2009-06-11 10::56:38] rmatte: chudler: doesn't really help when you're applying the templates against multiple devices with different usernames and passwords set for WMI in their zproperties
[2009-06-11 10::56:42] rocket: mistich: eg remove_holt.py rrd_file
[2009-06-11 10::57:06] ** kgoedtel joined the chat room.
[2009-06-11 10::57:19] mistich: have a question for everyone how are you handling spikes in graphs
[2009-06-11 10::57:25] cgibbons: that's certainly why we have a daemon and WinPerf datatype in enterpise, so it can manage all that without having to spawn commands
[2009-06-11 10::57:26] chudler: rmatte: good point. what might the code look like that can censor the password? wouldn't it need to know the password? or just /-p .*? -/ ?
[2009-06-11 10::57:43] cgibbons: IMO, using a command script for monitoring is a short-term quick-fix work-around, and a real ZenPack should do better
[2009-06-11 10::57:43] mistich: min and max I hate to use it because you have to recreate the rrd
[2009-06-11 10::57:54] bwest: Another threshold questons. Right now there is the min max, but if I want ok warning critical so 3 levels how do i accomplish this
[2009-06-11 10::58:00] cgibbons: (i'm also looking in our change history to find out if we hav an example of passing in the password via stdin)
[2009-06-11 10::58:21] rocket: bwest: can you use better grammar your question is hard to understand
[2009-06-11 10::58:35] bwest: Sorry
[2009-06-11 10::58:52] ** tristil left the chat room.
[2009-06-11 10::58:57] bwest: Using thresholds there is a min and max
[2009-06-11 10::58:59] rmatte: chudler: Zenoss should already know the password since it's passing it to the command, so when the command fails, identify that that is a password value and only display the rest of the command.
[2009-06-11 10::59:17] sergeymasushko: bwest: can you add 2 different minmax thresholds to the same datapoint?
[2009-06-11 10::59:20] bwest: but I want a third level so I can warn as the threshold is approachong max
[2009-06-11 10::59:52] mistich: thats what we do
[2009-06-11 11::00:17] ** Guest52519 joined the chat room.
[2009-06-11 11::00:50] Guest52519: hmm
[2009-06-11 11::00:54] rmatte: cgibbons: thanks, if that can be done then that would be a viable solution
[2009-06-11 11::01:16] ** Guest52519 is now known as euphoria.
[2009-06-11 11::01:39] rmatte: as a side note, I'm finding myself writing quite a lot of command based scripts for monitoring all sorts of different stuff, such as the stock windows snmp agent which is a total piece of garbage and likes to move SNMP values around a lot
[2009-06-11 11::01:39] ** tristil joined the chat room.
[2009-06-11 11::01:45] rocket: bwest: there is a new community threshold zenpack
[2009-06-11 11::02:11] cgibbons: yeah monitoring Windows with SNMP is problematic even with the nice commerical agents
[2009-06-11 11::02:12] rocket: rangethreshold = warn between 3 and 4
[2009-06-11 11::02:20] jsynack: rmatte: move things around or put them in a table with different iid's ?
[2009-06-11 11::02:23] mrayzenoss: http://www.zenoss.com/community/projects/zenpacks/rangethreshold
[2009-06-11 11::02:28] rmatte: they have the snmp value for the RAM stored below the SNMP values for the physical drives, so the value to poll RAM is different on every server, the CPU OIDs tend to wander up and down, one day the first CPU will be .1, then it'll be .2, so I have to use scripts to deal with that
[2009-06-11 11::02:31] rocket: minmax threshold critical > 4
[2009-06-11 11::02:32] bwest: even better solution
[2009-06-11 11::03:39] ** lukecyca joined the chat room.
[2009-06-11 11::03:49] rmatte: my scripts are working quite nicely though, also allows me to easily calculate percentages and stuff
[2009-06-11 11::04:05] ** euphoria is now known as myeuphoria.
[2009-06-11 11::05:15] Jontler: i have a question about default performance monitoring of the cpu
[2009-06-11 11::05:49] rocket: bwest: make sense?
[2009-06-11 11::05:54] Jontler: i have multi-core machines, though - is that cpu graph a compilation of the cpus, is it the first cpu only, or is it something else?
[2009-06-11 11::06:08] bwest: yes perfect sense
[2009-06-11 11::06:10] rocket: Jontler: from snmp its probably an average
[2009-06-11 11::06:15] rmatte: Jontler: WMI or SNMP?
[2009-06-11 11::06:16] Jontler: ssh
[2009-06-11 11::06:25] rocket: bwest I thing range is <=
[2009-06-11 11::06:33] rmatte: ssh would be a total I believe
[2009-06-11 11::06:35] Jontler: im monitoring using ssh and the zenoss plugins on the machine side
[2009-06-11 11::06:41] rocket: bwest: minmax is >=
[2009-06-11 11::07:40] bwest: on one of my windows Servers the SNMP service will not stay on. It Starts and then a pop up comes up that says some sSNMP
[2009-06-11 11::08:14] ** chudler left the chat room.
[2009-06-11 11::08:25] ** kisielk joined the chat room.
[2009-06-11 11::08:27] ** jsynack left the chat room.
[2009-06-11 11::08:27] bwest: how do I ovveride this so SNMP stays running
[2009-06-11 11::08:36] rmatte: Jontler: easy way to find out, do "cat /dev/urandom > /dev/null", that should eat up all of the CPU on the first core, if it doesn't target the first core try again until it does, then check Zenoss in a bit for the current value, if it's 100% it's just monitoring that one core, if it's 25%ish, then it's monitoring everything
[2009-06-11 11::08:55] ** jplouis left the chat room.
[2009-06-11 11::09:19] Jontler: i will try that rmatte, thanks for the suggestion
[2009-06-11 11::09:23] rmatte: np
[2009-06-11 11::10:10] Jontler: ive never heard of it
[2009-06-11 11::10:18] rmatte: you'll like it
[2009-06-11 11::10:21] etank: htop is da`bomb
[2009-06-11 11::11:03] Jontler: oh hotness
[2009-06-11 11::11:09] rmatte: hehe
[2009-06-11 11::11:43] Jontler: yes it is
[2009-06-11 11::11:44] etank: si
[2009-06-11 11::11:50] mrayzenoss: wow, I didn't know top could be so sexxy
[2009-06-11 11::11:59] etank: i like the screen-profiles stuff that ubuntu has
[2009-06-11 11::12:09] rmatte: you can actually use mouse clicks with htop btw
[2009-06-11 11::12:36] Jontler: i would never have tried
[2009-06-11 11::13:05] rmatte: yessir
[2009-06-11 11::13:37] Jontler: that's much more intuitive
[2009-06-11 11::14:09] rmatte: but yeh, run that command that I told you to run and you should see the first core kick up to 100
[2009-06-11 11::14:21] Jontler: it's already rolling
[2009-06-11 11::14:25] rmatte: cool
[2009-06-11 11::14:31] Jontler: i didnt realize that would only tank one core
[2009-06-11 11::14:37] bwest: Another quick SNMP Question. on one of my windows servers, the SNMP service will not stay on. I get an message saying the SNMP service on your local machine has started and stopped. This happes when the service has nothing to do. Is there anyway I can fix this?
[2009-06-11 11::15:54] rmatte: bwest: never seen that behaviour before
[2009-06-11 11::15:58] sergeymasushko: heh... so no one can't help me with the graphs?
[2009-06-11 11::16:13] rmatte: Jontler: yeh, it's a single thread process so it just tanks 1
[2009-06-11 11::16:57] mrayzenoss: sergeymasushko: the links to the images are bad
[2009-06-11 11::17:09] bwest: very strange I know, on that server I cannot get snmp to work
[2009-06-11 11::18:39] rmatte: cgibbons: any luck finding that stdin method info?
[2009-06-11 11::20:12] cgibbons: don't see an example yet. looks like we just stopped using command for anything like that.
[2009-06-11 11::20:34] rmatte: ah
[2009-06-11 11::20:42] mistich: well you all have a good day
[2009-06-11 11::20:45] ** mistich left the chat room.
[2009-06-11 11::21:00] rmatte: I'll hunt around and see if I can find some method of doing it
[2009-06-11 11::21:49] mrayzenoss: BTW, anyone looking for a Zenoss job in the bay area, let me know
[2009-06-11 11::22:02] Jontler: rmatte: so the percentage is scaled from 0 to 200%, almost 200% was idle before, and zero is idle now
[2009-06-11 11::23:07] ckrough: if its gover over 100% I would think its adding all the cores
[2009-06-11 11::23:33] sergeymasushko: mrayzenoss: I've fixed the links.
[2009-06-11 11::23:43] rmatte: yeh, it might be adding them up
[2009-06-11 11::23:53] Jontler: 2
[2009-06-11 11::25:00] rmatte: yeh, then something is definitely messed
[2009-06-11 11::25:09] Jontler: oh wait
[2009-06-11 11::25:11] mrayzenoss: we're going to wrap this up, I'll let cgibbons get back to closing tickets. Keep the ZenPacks coming in and stop by at SouthEast LinuxFest if you're in the neighborhood
[2009-06-11 11::25:15] rmatte: keep firing up commands until you see some action
[2009-06-11 11::25:17] Jontler: something is messed alright, and it's called my eyes
[2009-06-11 11::25:24] rmatte: run 2 or 3 in parallel
[2009-06-11 11::25:32] Jontler: i zoomed in on the graph
[2009-06-11 11::25:33] rmatte: need glasses do you?
[2009-06-11 11::25:35] ** You changed the topic to "Zenoss is at SouthEast LinuxFest this weekend".
[2009-06-11 11::25:40] rmatte:
[2009-06-11 11::25:48] Jontler: well i was all the way zoomed out and it looked like it was dropped off completely
[2009-06-11 11::26:28] rmatte: yup
[2009-06-11 11::26:59] Jontler: well, thanks for helping me solve that mystery
[2009-06-11 11::27:04] rmatte: no problem
[2009-06-11 11::29:06] mrayzenoss: every 2 weeks, Thursday
[2009-06-11 11::29:21] rmatte: ah, I thought it only happened once a month
[2009-06-11 11::29:24] mrayzenoss: http://www.zenoss.com/community/discuss/irc
[2009-06-11 10::03:02] cgibbons: *golfclap*
[2009-06-11 10::03:10] mrayzenoss: good morning/afternoon/evening everyone, cgibbons and I will be here to answer your questions
[2009-06-11 10::03:45] rmatte: good morning Matt
[2009-06-11 10::03:50] cgibbons: bwest: yeah, rocket's right. it's much better to use the built-in WMI support to do your own modeling plugins if you need.
[2009-06-11 10::04:51] bwest: I am using 2.4.1
[2009-06-11 10::06:39] cgibbons: not a good way to do it, really. winexe isn't a good tool. zenoss ships with wmic as a command-line tool that you can use, plus you can run WMI queries directly from the python code.
[2009-06-11 10::06:59] ** jplouis joined the chat room.
[2009-06-11 10::08:15] bwest: So how do I get to the command line when modeling to use wmic
[2009-06-11 10::08:21] rmatte: ok, so my main question for the developers: On the OS tab, port status appears to be determined during modelling. Why is no polling done on monitored interfaces each polling cycle to check status? I'm having to use a threshold to properly monitor administrative status on ports, and it doesn't update the status on the OS tab. Ports stay showing as up when they should be showing as down, and vice versa.
[2009-06-11 10::08:45] cgibbons: for example I just did this here: wmic -U 'zenoss\administrator' //hostname 'select SerialNumber from Win32_SystemEnclosure'
[2009-06-11 10::10:10] ** magnachef_ left the chat room.
[2009-06-11 10::10:13] mrayzenoss: cgibbons: this one? http://www.zenoss.com/community/projects/zenpacks/windowswmidevicetemplate
[2009-06-11 10::10:25] cgibbons: look at WMIPlugin.py in Products.ZenWin for somethinng you can extend in your own modeling plugins to do WMI queries
[2009-06-11 10::10:41] bwest: ok thx
[2009-06-11 10::11:58] cgibbons: the way WMIPlugin.py is done you don't have to worry about executing the query yourself, you just override the queries() method with the WMI queries you want it to execute
[2009-06-11 10::13:07] rmatte: cgibbons: ok, is it something that's being looked at for the future at all?
[2009-06-11 10::14:07] cgibbons: it hasn't been escalated to the point where the enterprise customers have asked for it (surprisingly). i'm surprised it doesn't come up more.
[2009-06-11 10::14:31] ckrough: Not very manageable on a large scale.
[2009-06-11 10::17:01] ** sergeymasushko left the chat room.
[2009-06-11 10::18:07] bwest: If I have an environment that has about 350 components it's monitoring, running 2.4.1 on linux what server requirments shold be large enough? The components are 200 servers and 150 switches and routers
[2009-06-11 10::18:24] rmatte: I'm just looking at this from the perspective of Zenoss' campaign to win over people from the other larger monitoring companies/tools, but if you're lacking basic functionality like this, it doesn't really make a good showing.
[2009-06-11 10::19:48] cgibbons: there's a chart we have on teh website with requirements... let me find it
[2009-06-11 10::20:19] ckrough: bwest: depents on the number of things you are collecting, how long you keep the rrd data, and how frequently you collect it
[2009-06-11 10::21:01] bwest: they are interested in baselining and they are monitoring this data around the clock
[2009-06-11 10::22:56] mrayzenoss: make sure you're on a fast filesystem
[2009-06-11 10::23:01] cgibbons: mattray, do you have any stats available with some of our really big customers?
[2009-06-11 10::23:15] mrayzenoss: well, the largest community install 1 box is 8500 devices
[2009-06-11 10::23:26] ckrough: 1 box?
[2009-06-11 10::23:36] mrayzenoss: yeah, 8-way 32 gigs of RAM
[2009-06-11 10::23:54] rocket: ckrough: devices isnt really indicitive of the amount of data they are collecting
[2009-06-11 10::23:54] mrayzenoss: someone else is doing 10K
[2009-06-11 10::24:02] ckrough: I can only get 1600 devices on that spec, with 6 15k RAID10 storage config
[2009-06-11 10::24:08] rocket: ckrough: need to look at the # of datapoints collected
[2009-06-11 10::24:19] ckrough: 75k OIDs per box on that spec
[2009-06-11 10::25:01] cgibbons: I suspect a lot of the large upcoming installations will be doing mostly Windows devices using WMI/PerfMon, so that'll change the game some
[2009-06-11 10::25:32] mrayzenoss: if you go to Collectors->localhost->Performance you can see your stats
[2009-06-11 10::27:05] bwest: one collector with 10k
[2009-06-11 10::27:19] rmatte: one collector with 10k is insane
[2009-06-11 10::27:27] ckrough: 10k what
[2009-06-11 10::27:34] Jontler: devices
[2009-06-11 10::27:37] rmatte: I assumed he meant devices
[2009-06-11 10::27:41] mrayzenoss: yeah
[2009-06-11 10::27:52] rmatte: what kind of devices though?
[2009-06-11 10::27:59] mrayzenoss: cable modems?
[2009-06-11 10::28:14] bwest: no not devices data points
[2009-06-11 10::28:14] rmatte: oh, that's not really incredibly impressive then
[2009-06-11 10::28:24] mrayzenoss: right
[2009-06-11 10::28:44] ckrough: disk IO seems to be the killer. anyone runnign this on SSD yet?
[2009-06-11 10::29:41] cgibbons: it scales nicely horizontally for that issue, since data (sans events) is kept on each collector.
[2009-06-11 10::29:45] rmatte: I'm polling like 18.7k datapoints on my largest install so far, but I'm going to have much larger projects coming down the pipe soon
[2009-06-11 10::30:27] bwest: but for 10K datapoints I should just look at the sizing chart?
[2009-06-11 10::30:53] mrayzenoss: here's the bragging thread for Core installs: http://forums.zenoss.com/viewtopic.php?t=6230
[2009-06-11 10::31:24] ckrough: >430k OID on my biggest install
[2009-06-11 10::31:45] ** mistich joined the chat room.
[2009-06-11 10::31:51] ** willwh joined the chat room.
[2009-06-11 10::31:53] mrayzenoss: heh, I'm doing about 110 on my home install
[2009-06-11 10::33:00] cgibbons: hrmph only 220 on my home setup, which i haven't looked @ in months
[2009-06-11 10::33:25] ckrough: Yeah, I plan to set it up at home to watch my total usage with NetFlix... b/w caps are coming...
[2009-06-11 10::33:30] cgibbons: doh it's only 2.3.1, bad me
[2009-06-11 10::33:48] rocket: cgibbons: heh I dont have as many datapoints .. but a fairly large number of zenpacks installed
[2009-06-11 10::33:59] mrayzenoss: my home install is where the source build on PPC gets tested
[2009-06-11 10::34:05] ** chudler joined the chat room.
[2009-06-11 10::34:11] cgibbons: you still have that powerbook??
[2009-06-11 10::34:18] mistich: does anyone know the status of IPV6 in zenoss
[2009-06-11 10::34:35] mrayzenoss: mistich: it hasn't gotten prioritized very high
[2009-06-11 10::34:44] ckrough: plays sad trombone
[2009-06-11 10::34:46] mrayzenoss: mistich: there is a HOWTO somewhere for doing IPV6 ping
[2009-06-11 10::35:24] bwest: I am trying to model SIP adapters. SNMP is not available on them, but they can be identified with Mac addresses, can zenoss get this information
[2009-06-11 10::35:25] mistich: yes but what about snmp ssh etc
[2009-06-11 10::35:30] mrayzenoss: I did get some emails from a university in China that said their class project was to port Zenoss to IPV6
[2009-06-11 10::35:54] ckrough: Is there much demand for making reporting more robust? Maybe moving graph definitions to re-usable objects, etc
[2009-06-11 10::36:01] mistich: yes I saw that too has anyone followed up with them
[2009-06-11 10::36:06] mrayzenoss: I did
[2009-06-11 10::36:36] ** sergeymasushko joined the chat room.
[2009-06-11 10::36:38] cgibbons: there's some good porting enhancements coming in the next release actually
[2009-06-11 10::36:43] mrayzenoss: ckrough: there are already some changes going into King Crab for reporting
[2009-06-11 10::36:50] rmatte: mrayzenoss: probably, I'm sure they found out that it was much more complicated than they initially thought
[2009-06-11 10::36:57] mrayzenoss: rmatte: definitely
[2009-06-11 10::37:05] ckrough: mrayzenoss: I will send you swag if you do
[2009-06-11 10::37:09] mrayzenoss: heh
[2009-06-11 10::37:36] rmatte: you'd almost have to have some sort of layer between Zenoss and the device that takes a dummy ipv4 address and forwards the traffic to an ipv6 address and vice versa
[2009-06-11 10::37:50] mrayzenoss: yeah, it's non-trivial
[2009-06-11 10::37:57] mistich: next question moving away from rrds to a relational database
[2009-06-11 10::38:03] ckrough: ^^ this
[2009-06-11 10::38:11] sergeymasushko: can someone help me with the graph issue. I believe it should be simple, but I stack on it. http://forums.zenoss.com/viewtopic.php?t=10008
[2009-06-11 10::38:32] rmatte: Matt, any progress on figuring out what the cause of that zenpack load time issue could be?
[2009-06-11 10::39:07] mrayzenoss: rmatte: jplouis was looking at it, I believe nothing obvious was showing up yet
[2009-06-11 10::39:35] cgibbons: meaning you want to export your raw RRD data into a relational database for custom reporting, mistich, or you want to change the data collection to store in a database directly?
[2009-06-11 10::39:36] rmatte: cool
[2009-06-11 10::39:44] mrayzenoss: mistich: the idea of moving off of RRD does come up, just not sure what we'd move to
[2009-06-11 10::40:12] rmatte: one of our guys here came up with a way to dump an RRD to XML and then pull from there
[2009-06-11 10::40:16] mistich: well for reporting relational database would be nice
[2009-06-11 10::40:35] bwest: I am sorry i am so full of questions, but is it possible to send mysql queries from zenoss to read the amount of data in a table and create events off this?
[2009-06-11 10::40:36] mistich: but want realtime data
[2009-06-11 10::40:37] jplouis: zenpack slowness has to do with the way the relationships are loaded. The zenpack has a relationship to each packable
[2009-06-11 10::40:43] mrayzenoss: bwest: yes
[2009-06-11 10::41:10] cgibbons: I can tell you from experience that dumping that much raw data into a relational database in real-time is not a fun path.
[2009-06-11 10::41:54] mistich: then once in a relational database reporting becomes very easily can use any package to do reporting
[2009-06-11 10::41:54] mrayzenoss: bwest: have you tried the Core MySQL ZenPack?
[2009-06-11 10::42:03] rmatte: bwest, you'd have to write a script for that and use it as a command based datapoint
[2009-06-11 10::42:17] bwest: perfect
[2009-06-11 10::42:18] mrayzenoss: bwest: I just got another massive set of ZenPacks for doing databases yesterday (MySQL, PostgreSQL, MS SQL, ZeoDB)
[2009-06-11 10::42:24] rmatte: although that MySQL ZenPack might be able to do it, haven't looked at it
[2009-06-11 10::42:37] mistich: oh I understand my install is huge but everyday getting asked for more and more reports
[2009-06-11 10::42:40] mrayzenoss: it provides some stats, it may be good enough
[2009-06-11 10::43:02] rmatte: MS SQL you say?
[2009-06-11 10::43:07] mrayzenoss: rmatte: yeah
[2009-06-11 10::43:18] rmatte: I've been waiting for a good community Zenpack for that
[2009-06-11 10::43:28] mrayzenoss: don't like the existing one?
[2009-06-11 10::43:41] bwest: when will that one be published My SQL
[2009-06-11 10::43:57] mrayzenoss: probably over the weekend
[2009-06-11 10::44:22] bwest: excellent, your contest is working
[2009-06-11 10::44:30] ** CamargoBP joined the chat room.
[2009-06-11 10::44:32] mrayzenoss: yeah, we already have the same number of entries as last year
[2009-06-11 10::44:53] mistich: yeah we have about 6 zenpacks one day well publish
[2009-06-11 10::45:07] chudler: I do have at least 4 more to send you, but I have to get time to polish them
[2009-06-11 10::45:10] mistich: well I'll wait till the prizes are bumped
[2009-06-11 10::45:53] mrayzenoss: the current backlog is: PredictiveThreshold, AIX SNMP, Netware, LinuxMonitorAddon, Gentoo, SuSE, Jabber XMMP, RDBMS, MySQL, MS SQL Server, PostgreSQL, ZeoDB
[2009-06-11 10::46:27] mistich: PredictiveThreshold is this holt winters
[2009-06-11 10::46:32] mrayzenoss: yes
[2009-06-11 10::46:40] mistich: already have that one
[2009-06-11 10::46:47] mrayzenoss: I'm comparing the Enterprise one with the Community one now
[2009-06-11 10::47:25] bwest: Are there any good zenpacks to monitor VOIP in core or enterprise?
[2009-06-11 10::47:27] etank: mrayzenoss: i think that the biggest issue for me was getting the Perfmon ZenPack to work
[2009-06-11 10::48:00] mrayzenoss: bwest: the Asterisk one seems to have some issues, but you could probably adapt any Nagios plugin that already exists
[2009-06-11 10::48:05] cgibbons: i may be biased, but if it were me, if a zenpack uses winexe I'd ditch it and never look bck.
[2009-06-11 10::48:06] sergeymasushko: bwest: what do you want to monitor in the VOIP?
[2009-06-11 10::48:42] bwest: Right now they use Cricket to monitor calls per minute users and the such
[2009-06-11 10::49:01] etank: mrayzenoss: is there a PredictiveThreshold community Pack? or just for Enterprise?
[2009-06-11 10::49:13] mrayzenoss: this will be for Core
[2009-06-11 10::49:35] etank: really? how different would it be from the Enterprise one?
[2009-06-11 10::49:50] mistich: what is the difference between them?
[2009-06-11 10::49:53] mrayzenoss: I just set it up yesterday, they take a few days to generate data
[2009-06-11 10::49:58] ** jsynack joined the chat room.
[2009-06-11 10::50:07] mrayzenoss: I've got both of them installed and testing
[2009-06-11 10::50:17] sergeymasushko: bwest: what is the voip platform (sip server ?)
[2009-06-11 10::50:45] bwest: yes
[2009-06-11 10::50:53] rmatte: I have another question... when executing a command based script where you need to pass a password value to it, if the command fails the password gets spit out in plaintext in the command failed event. I logged a ticket for this and was told to "See the ZenPack creator for a fix". I've already tried fixing this ZenPack by using a python script to pull the username and password values from the Zope database, but after extensive testing I have determined that
[2009-06-11 10::50:59] mistich: one thing to look for is low data in the kb holtwinters doesn't like it and sends false alerts
[2009-06-11 10::51:15] rmatte: The only way to possibly fix this security issue is to have Zenoss intelligently sensor password values in command failed events.
[2009-06-11 10::52:04] mrayzenoss: well, I believe the ZenPack in question was one we don't even have access to
[2009-06-11 10::52:25] cgibbons: the reason that was given as the answer rmatte is because there are two ways to deal with it: 1) not use COMMAND datatypes (which is what we mostly do), 2) a command script that needs a password gets it sent via stdin instead of on the command-line
[2009-06-11 10::52:25] mrayzenoss: it's one of James Dastrup's, which we haven't looked at since he hasn't sent it to us for hosting
[2009-06-11 10::52:46] sergeymasushko: bwest: for monitoring sip servers I use a shell command that run SIPP util with the different scenarios... then parse result and make an event if its needed. also I'm monitoring active calls on my cisco AS5300 GWs...
[2009-06-11 10::52:49] rocket: mistich: I wrote the predictive zenpack .. and tried to make it as configurable as possible
[2009-06-11 10::53:12] rmatte: yeh, but it's not specific to that zenpack is my point, this issue pops up with any type of command based poll that needs a password passed to it from Zenoss
[2009-06-11 10::53:21] mistich: rocket: nice
[2009-06-11 10::53:24] rmatte: whether it be a custom built script, or a core Zenpack
[2009-06-11 10::53:29] bwest: serg: sounds interesting, is there a way for you to share more with me
[2009-06-11 10::53:29] cgibbons: there's a mechanism to do it that doesn't require it being sent on the command-line
[2009-06-11 10::54:07] rmatte: which would be using a python script to pull the values directly from Zope, correct?
[2009-06-11 10::54:10] mistich: like that it retofits the rrd but does it also remove the holtwinters if you want to remove it
[2009-06-11 10::54:11] rocket: mistich: so if you are getting false alerts my threshold will also allow you to go and reapply the beta and gamma and window etc
[2009-06-11 10::54:12] sergeymasushko: bwest: a bit busy for now... lets talk about a bit later...
[2009-06-11 10::54:25] bwest: ok
[2009-06-11 10::54:25] rocket: mistich: it does not currently remove
[2009-06-11 10::54:31] mistich: can't wait to look at it
[2009-06-11 10::54:54] rocket: mistich: as I cant think of a good way to have zenoss do that on removal ..
[2009-06-11 10::55:22] rmatte: I've already done that and modified the scripts in 3 different WMI zenpacks to use it, but we had a Zenoss server crash because the retrieval of the password and username values for each poll was way too intensive, if you can suggest a more efficient way of doing it then I'd be happy to modify the Zenpacks in that fashion.
[2009-06-11 10::55:46] mistich: rocket: have run into that before thats why I bring it up
[2009-06-11 10::56:05] chudler: when I need a password, I put it in a file, and read it from there. Not sure how/if that helps you, rmatte
[2009-06-11 10::56:14] rmatte: It's either a little bit of code to sensor the password, or large loss of performance in the scripts
[2009-06-11 10::56:17] rocket: mistich: it might be something that I could make a command line to do
[2009-06-11 10::56:38] rmatte: chudler: doesn't really help when you're applying the templates against multiple devices with different usernames and passwords set for WMI in their zproperties
[2009-06-11 10::56:42] rocket: mistich: eg remove_holt.py rrd_file
[2009-06-11 10::57:06] ** kgoedtel joined the chat room.
[2009-06-11 10::57:19] mistich: have a question for everyone how are you handling spikes in graphs
[2009-06-11 10::57:25] cgibbons: that's certainly why we have a daemon and WinPerf datatype in enterpise, so it can manage all that without having to spawn commands
[2009-06-11 10::57:26] chudler: rmatte: good point. what might the code look like that can censor the password? wouldn't it need to know the password? or just /-p .*? -/ ?
[2009-06-11 10::57:43] cgibbons: IMO, using a command script for monitoring is a short-term quick-fix work-around, and a real ZenPack should do better
[2009-06-11 10::57:43] mistich: min and max I hate to use it because you have to recreate the rrd
[2009-06-11 10::57:54] bwest: Another threshold questons. Right now there is the min max, but if I want ok warning critical so 3 levels how do i accomplish this
[2009-06-11 10::58:00] cgibbons: (i'm also looking in our change history to find out if we hav an example of passing in the password via stdin)
[2009-06-11 10::58:21] rocket: bwest: can you use better grammar your question is hard to understand
[2009-06-11 10::58:35] bwest: Sorry
[2009-06-11 10::58:52] ** tristil left the chat room.
[2009-06-11 10::58:57] bwest: Using thresholds there is a min and max
[2009-06-11 10::58:59] rmatte: chudler: Zenoss should already know the password since it's passing it to the command, so when the command fails, identify that that is a password value and only display the rest of the command.
[2009-06-11 10::59:17] sergeymasushko: bwest: can you add 2 different minmax thresholds to the same datapoint?
[2009-06-11 10::59:20] bwest: but I want a third level so I can warn as the threshold is approachong max
[2009-06-11 10::59:52] mistich: thats what we do
[2009-06-11 11::00:17] ** Guest52519 joined the chat room.
[2009-06-11 11::00:50] Guest52519: hmm
[2009-06-11 11::00:54] rmatte: cgibbons: thanks, if that can be done then that would be a viable solution
[2009-06-11 11::01:16] ** Guest52519 is now known as euphoria.
[2009-06-11 11::01:39] rmatte: as a side note, I'm finding myself writing quite a lot of command based scripts for monitoring all sorts of different stuff, such as the stock windows snmp agent which is a total piece of garbage and likes to move SNMP values around a lot
[2009-06-11 11::01:39] ** tristil joined the chat room.
[2009-06-11 11::01:45] rocket: bwest: there is a new community threshold zenpack
[2009-06-11 11::02:11] cgibbons: yeah monitoring Windows with SNMP is problematic even with the nice commerical agents
[2009-06-11 11::02:12] rocket: rangethreshold = warn between 3 and 4
[2009-06-11 11::02:20] jsynack: rmatte: move things around or put them in a table with different iid's ?
[2009-06-11 11::02:23] mrayzenoss: http://www.zenoss.com/community/projects/zenpacks/rangethreshold
[2009-06-11 11::02:28] rmatte: they have the snmp value for the RAM stored below the SNMP values for the physical drives, so the value to poll RAM is different on every server, the CPU OIDs tend to wander up and down, one day the first CPU will be .1, then it'll be .2, so I have to use scripts to deal with that
[2009-06-11 11::02:31] rocket: minmax threshold critical > 4
[2009-06-11 11::02:32] bwest: even better solution
[2009-06-11 11::03:39] ** lukecyca joined the chat room.
[2009-06-11 11::03:49] rmatte: my scripts are working quite nicely though, also allows me to easily calculate percentages and stuff
[2009-06-11 11::04:05] ** euphoria is now known as myeuphoria.
[2009-06-11 11::05:15] Jontler: i have a question about default performance monitoring of the cpu
[2009-06-11 11::05:49] rocket: bwest: make sense?
[2009-06-11 11::05:54] Jontler: i have multi-core machines, though - is that cpu graph a compilation of the cpus, is it the first cpu only, or is it something else?
[2009-06-11 11::06:08] bwest: yes perfect sense
[2009-06-11 11::06:10] rocket: Jontler: from snmp its probably an average
[2009-06-11 11::06:15] rmatte: Jontler: WMI or SNMP?
[2009-06-11 11::06:16] Jontler: ssh
[2009-06-11 11::06:25] rocket: bwest I thing range is <=
[2009-06-11 11::06:33] rmatte: ssh would be a total I believe
[2009-06-11 11::06:35] Jontler: im monitoring using ssh and the zenoss plugins on the machine side
[2009-06-11 11::06:41] rocket: bwest: minmax is >=
[2009-06-11 11::07:40] bwest: on one of my windows Servers the SNMP service will not stay on. It Starts and then a pop up comes up that says some sSNMP
[2009-06-11 11::08:14] ** chudler left the chat room.
[2009-06-11 11::08:25] ** kisielk joined the chat room.
[2009-06-11 11::08:27] ** jsynack left the chat room.
[2009-06-11 11::08:27] bwest: how do I ovveride this so SNMP stays running
[2009-06-11 11::08:36] rmatte: Jontler: easy way to find out, do "cat /dev/urandom > /dev/null", that should eat up all of the CPU on the first core, if it doesn't target the first core try again until it does, then check Zenoss in a bit for the current value, if it's 100% it's just monitoring that one core, if it's 25%ish, then it's monitoring everything
[2009-06-11 11::08:55] ** jplouis left the chat room.
[2009-06-11 11::09:19] Jontler: i will try that rmatte, thanks for the suggestion
[2009-06-11 11::09:23] rmatte: np
[2009-06-11 11::10:10] Jontler: ive never heard of it
[2009-06-11 11::10:18] rmatte: you'll like it
[2009-06-11 11::10:21] etank: htop is da`bomb
[2009-06-11 11::11:03] Jontler: oh hotness
[2009-06-11 11::11:09] rmatte: hehe
[2009-06-11 11::11:43] Jontler: yes it is
[2009-06-11 11::11:44] etank: si
[2009-06-11 11::11:50] mrayzenoss: wow, I didn't know top could be so sexxy
[2009-06-11 11::11:59] etank: i like the screen-profiles stuff that ubuntu has
[2009-06-11 11::12:09] rmatte: you can actually use mouse clicks with htop btw
[2009-06-11 11::12:36] Jontler: i would never have tried
[2009-06-11 11::13:05] rmatte: yessir
[2009-06-11 11::13:37] Jontler: that's much more intuitive
[2009-06-11 11::14:09] rmatte: but yeh, run that command that I told you to run and you should see the first core kick up to 100
[2009-06-11 11::14:21] Jontler: it's already rolling
[2009-06-11 11::14:25] rmatte: cool
[2009-06-11 11::14:31] Jontler: i didnt realize that would only tank one core
[2009-06-11 11::14:37] bwest: Another quick SNMP Question. on one of my windows servers, the SNMP service will not stay on. I get an message saying the SNMP service on your local machine has started and stopped. This happes when the service has nothing to do. Is there anyway I can fix this?
[2009-06-11 11::15:54] rmatte: bwest: never seen that behaviour before
[2009-06-11 11::15:58] sergeymasushko: heh... so no one can't help me with the graphs?
[2009-06-11 11::16:13] rmatte: Jontler: yeh, it's a single thread process so it just tanks 1
[2009-06-11 11::16:57] mrayzenoss: sergeymasushko: the links to the images are bad
[2009-06-11 11::17:09] bwest: very strange I know, on that server I cannot get snmp to work
[2009-06-11 11::18:39] rmatte: cgibbons: any luck finding that stdin method info?
[2009-06-11 11::20:12] cgibbons: don't see an example yet. looks like we just stopped using command for anything like that.
[2009-06-11 11::20:34] rmatte: ah
[2009-06-11 11::20:42] mistich: well you all have a good day
[2009-06-11 11::20:45] ** mistich left the chat room.
[2009-06-11 11::21:00] rmatte: I'll hunt around and see if I can find some method of doing it
[2009-06-11 11::21:49] mrayzenoss: BTW, anyone looking for a Zenoss job in the bay area, let me know
[2009-06-11 11::22:02] Jontler: rmatte: so the percentage is scaled from 0 to 200%, almost 200% was idle before, and zero is idle now
[2009-06-11 11::23:07] ckrough: if its gover over 100% I would think its adding all the cores
[2009-06-11 11::23:33] sergeymasushko: mrayzenoss: I've fixed the links.
[2009-06-11 11::23:43] rmatte: yeh, it might be adding them up
[2009-06-11 11::23:53] Jontler: 2
[2009-06-11 11::25:00] rmatte: yeh, then something is definitely messed
[2009-06-11 11::25:09] Jontler: oh wait
[2009-06-11 11::25:11] mrayzenoss: we're going to wrap this up, I'll let cgibbons get back to closing tickets. Keep the ZenPacks coming in and stop by at SouthEast LinuxFest if you're in the neighborhood
[2009-06-11 11::25:15] rmatte: keep firing up commands until you see some action
[2009-06-11 11::25:17] Jontler: something is messed alright, and it's called my eyes
[2009-06-11 11::25:24] rmatte: run 2 or 3 in parallel
[2009-06-11 11::25:32] Jontler: i zoomed in on the graph
[2009-06-11 11::25:33] rmatte: need glasses do you?
[2009-06-11 11::25:35] ** You changed the topic to "Zenoss is at SouthEast LinuxFest this weekend".
[2009-06-11 11::25:40] rmatte:
[2009-06-11 11::25:48] Jontler: well i was all the way zoomed out and it looked like it was dropped off completely
[2009-06-11 11::26:28] rmatte: yup
[2009-06-11 11::26:59] Jontler: well, thanks for helping me solve that mystery
[2009-06-11 11::27:04] rmatte: no problem
[2009-06-11 11::29:06] mrayzenoss: every 2 weeks, Thursday
[2009-06-11 11::29:21] rmatte: ah, I thought it only happened once a month
[2009-06-11 11::29:24] mrayzenoss: http://www.zenoss.com/community/discuss/irc