snuff-workmm.. having issue getting zabbix to recognise openBSD unixodbc..
snuff-workso.. openBSD uses '-liodbc' instead but even just changing that in the straight configure file (not the ac..) seems to still fail out
snuff-workok.. i seem to have found a way.. now to find how to put into the (at least it finds module now)
snuff-workmm.. with zabbix, is there any other way than using say 'ldd' to figure out what options have been compiled into the zabbix_proxy/zabbix_server binaries ?
zabbixbot[04ZBXNEXT-4258] 03Option to limit for Global notifications 14created by Kim Jongkwon14/JKKim (12
zabbixZBXNEXT-4258 Open Updated -1 days ago -- Option to limit for Global notifications
anoopjbIn zabbix_server.conf the maximum timeout for external script is 30secs. But my script takes more than 30s to execute, gets some files from a remote machine to Zabbix for further processing. Any options to increase timeout more than 30s
_ikke_anoopjb: the usual advise is to run that process in the background
_ikke_and either push the results to zabbix, or use an external script to fetch it
anoopjb_ikke_: I am printing some results(logs actually) and wanted it to be seen in zabbix. Is there a way to push the stdout to the external script item
yaoolhey, I am just wonder, I have a COU usage item for a host 'abc-host', now how can i find the history data for that.
yaoolWhat I want is get the CPU usage as per my host name..
q1xyaool: latest data screen, or do you need API?
zabbixbot[04ZBX-13117] 03Junctions and symlinks are not properly handled by vfs.dir.size on Windows 14created by Andris Zeila14/wiper (12
zabbixZBX-13117 Open Updated today -- Junctions and symlinks are not properly handled by vfs.dir.size on Windows
volteryaool`: Or use the search and "Latest data" from there.
Guest42681hi, I want import template from Simple way dont work: configuration - templates - import => browse file -> import
Guest42681Template is "Template Net Extreme Exos SNMPv2" from
Guest42681Import failed with "Cannot import template "Template Net Extreme EXOS SNMPv2", linked templates "Template Module Interfaces SNMPv2, Template Module EtherLike-MIB SNMPv2, Template Module Generic SNMPv2" do not exist."
Guest42681I'm nowbie with zabbix...
IngusHi, you need to download the missing templates first and then the last desired template.
IngusThe others should be in as well.
Guest42681Ingus: then I need download and import "Template Module Interfaces SNMPv2, Template Module EtherLike-MIB SNMPv2, Template Module Generic SNMPv2" and then try again with extreme, correct?
Guest42681Ingus: thankYou,
IngusYes, that is correct according to the error message you got.
Guest42681OK, I'm back :)
Guest42681Ingus: in I try search for "EtherLike-MIB SNMPv2" or simple "EtherLike-MIB" and without any result. Where may be my fault?
IngusDownload and import this first:
IngusIt says so in the description :)
Guest42681Ingus: my foult
al-geocan i create item that shows ping from host1 to host2, easyly? or i must create userparameter? zab server and agents 3.2
_ikke_You need to create a userparam, there is no existing agent item to ping another host
al-geothanks _ikke_
troffaskytriggers created by "Template Module Interfaces Simple SNMPv2" refer to Interface {#IFDESCR}: High bandwidth usage >{$IF_UTIL_MAX:"{#IFNAME}"}%
troffaskyi can guess at what the intention is here, but where is it documented?
_ikke_Where is what documented?
troffaskythe purpose of the trigger
_ikke_Other than warn about interfaces that are saturated?
troffaskyso i assume that it takes the interface speed and uses that as a reference point for the max bandwidith
troffaskyok, where is $IF_UTIL_MAX defined?
troffaskyis it derived from the link speed?
_ikke_That is a user macro
_ikke_{$..} are user macros
_ikke_you can define them on the template, host or globally
troffaskyok, that's good
_ikke_host > template > global
troffaskybecause in the cases where i actually care about link saturation, the L2 speed is not the max L3 speed
_ikke_right, this is mostly about L2 speed
troffaskyL2 speed is almost always 100M but L3 is capped somewhere below that depending on what has been provisioned on circuit by ISP
troffaskyso i want to override IF_UTIL_MAX per-interface
_ikke_note that that is a pecetnage
troffaskyah ok
IngusSome reading in this regard.
Richlvthat's one awesome documentation for templates
troffaskyso, ifName is "X1 (WAN)"
troffaskylink speed is 100M, provisioned bandwidth is 50M
troffaskyi need to override {$IF_UTIL_MAX:"X1 (WAN)"}
troffaskyie set it locally on the host
troffaskyi am going to set it to 50?
troffaskytriggers will fire when throughput hits 50Mbps for 15 minutes
_ikke_troffasky: depending on the device, it can also be possible to set a bandwith on the port itself
_ikke_troffasky: which you can read out
troffaskyyeah, no chance - Sonicwalls
_ikke_troffasky: what is the actual condition?
troffaskyit does actually work - i set that macro at 10 and it's firing because utilisation is over 20Mbps for an hour
troffaskymy problem is now merely a cosmetic one - 10M bandwidth on a 100M bearer, the alert will be at 10%, which doesn't "look" like a problem to the Zabbix operator
troffaskywhen in reality, link is 100% used
troffaskyalthough with 100% utilisation I expect the high ping time trigger to be firing :-)
_ikke_You could also set a value that you subtract from the link speed
troffaskyi am not quite clear what you mean?
_ikke_Add an additional macro that subtracts from the link speed. You set that macro to 0 by default so that it works when it's not explicitly set
troffaskycan i override the link speed with a user macro?
troffaskyeven in some cases, we have a gigabit mini switch between a pair of HA firewalls and demarc
troffaskywith perhaps 20M of provisioned bandwidth :-D
troffaskyi think i am going to use the factory trigger as the basis for my own and do it in Mbps not %
Guest42681I have configure host (switch) with working template "Template SNMP Device".
Guest42681I want add additional template "Template Net Extreme EXOS SNMPv2" to this host.
Guest42681On update I got following error "Two items ("Device contact details" and "Device contact details") cannot populate one host inventory field "Contact", this would lead to a conflict."
Guest42681For all is clear - that if before to add second template I 'Unlink and clear' first (already linked template) then I can add template without any errors. BUT, for today, I want learn something :-).
Guest42681Where and what I need to configure to avoid this dublicate error on to add both templates to host. For example, how to rename one of "Device contact details"?
troffaskyyou can't rename the inventory fields
binary106just had a great idea, a little javascript label could show you how much space your stat will take based on the history and update interval you set
volterACTION hires somebody
Guest42681troffasky: Ok, then I need configure template, correct?
q1x" * volter hires somebody" <- XD
q1xbecause that is what kapitalists do!
volterEverybody has a price!
golf4rmornin guys !
golf4rI have a weird issue... when I try to update items in a template , it gives me a blank screen after... I can see the top menu but the middle is blank and nothing happens, I have to refresh or hit back and my modifications are not taken into consideration...
golf4rbut it works with NOT templated items... I can update those
golf4rhave you seen something similar ??
troffaskymaybe PHP running out of memory
troffaskycheck apache logs
golf4ryeah I checked them
golf4rnothing implicit in it
golf4rmemory_limit = 256M
golf4rshould be enough ?
troffaskyif it wasn't, it would be logging about it
golf4rthen it is even more weird
troffaskyi suppose apache could be configured in such a way that this sort of thing doesn't get logged
golf4reven if I put 1024M I have the same behavior
zetheroo'The item is not discovered anymore and will be deleted in 9y 10m 13d' :D
zetherooOk, how can I change that to something more sensible?
troffaskyon the discovery rule for that item
Guest42681Q - how to speed-up screen making process? Is possible add all graphics from host to screen and then remove what not needed? Or something similar for lazy admin...
troffaskyalthough editing the discovery rule won't change an already-discovered item
zetheroo'Keep lost resources period (in days)' ?
zetherooit's set to 3600 :D
zetheroochanged it to 7
zetheroo'The item is not discovered anymore and will be deleted in 5d 22h 21m'
zetherooso it does seem to take effect on already-discovered items ...
troffaskyright that's interesteing
troffaskyi could have sworn it *did* do that, then i make changes to mine last week and it didn't
troffaskywhat version are you on?
Richlvtroffasky, entity changes do not, retention period affects them
Richlvas in, it's not an item property
troffaskyso...should editing the lost resource period affect already-discovered items, or not?
Richlvsure, it does
Richlvand is often used to clean things up manually, then reverted back
troffaskyhmm ok
troffaskywill do some more testing
golf4rOk ... about this memory issue, I think it is a bug but I want to get your input..
golf4rI have fixed my issue by opening /usr/share/zabbix/include/ and adding ini_set('memory_limit', '8192M'); at the top
golf4rnow I can update the template
golf4rnote that the template has more than 900 hosts associated to it
golf4rif I try to change memory value in php.ini ... even if I put 8192, it keeps doing the blank page when I update
golf4rbut if I add ini_set('memory_limit', '8192M'); at the top of, it works
golf4rand idea why?
Richlvyour php.ini is probably not used
Richlvhigh memory usage is a known issue and there are several issues about it
Richlvput a test php file in your zabbix directory and dump phpinfo()
Richlvsee which config file it uses
golf4rlet me check
golf4rLoaded Configuration File /etc/php.ini
golf4rso it is the right file I was changing memory settings in
Richlvis the changed value reflected ? and did you restart apache after making the change ?
golf4ryes it is and yes I did
golf4rI will test again
golf4rlocal value is 128M
golf4rand global value is 8192M
golf4roh yeah.. its overridden in zabbix.conf
golf4rok that was it.. but stilll! 512M to execute a simple update it seems like a huge memory leak
Richlvnot a leak, just... not efficient :>
Richlvthere are issues in jira, search for them
troffaskysomeone was talking about an epic leak here the other day...restarting zabbix-server once usage hit 100GB
golf4rwell anyway !
golf4rthanks ;)
_ikke_100GB mem?
finalbetaBrrr, configuring two proxy's with the same PSK name, but a differend PSK value is not great to debug :P
finalbetaThe funny thing is zabbix seems to rotate the keys, so half the connections from the proxy work, the other half fail.
Richlvtroffasky, that was server
Richlv_ikke_, yes
_ikke_Why do you have a zabbix server with that amount of memory?
_ikke_quite a large instance
troffaskyto work around the memory leak, i assume
troffaskymy largest server has 384GB so can't really just give a guest 100GB like that...
_ikke_our zabbix server has just 14G mem
troffaskymine a mere 8GB!
troffaskyand that's shared with unifi, alfresco, netxms, ...
golf4rI run on 16 GB
golf4rand 32GB for mysql db
_ikke_Some of our zabbix proxy server are a tiny bit oversized, but that's mostly due to hardware reuse
whosgonnaI'm looking to monitor a counter that resets when the host is restarted. My thought is to record the delta, since it should be an infrequently chagning value, and i really only care about when it increments. My initial thought was to monitor the delta, but won't this show large drops on reboot? i.e. if the counter value is at 200, and after reboot, it becomes zero, then the delta is minus 200. Is there a way to track only positive in
whosgonnacrements, etc?
troffaskyyou might be overthinking this whosgonna
troffaskyi don't see massive spikes on my interface graphs when a device reboots
whosgonnai could be ;)
troffaskyzabbix seems to handle overflow detection without any work on my part
whosgonnathat is an excellent point, but network traffic isn't just delta,it's an average of delta per second (i think...), so it may process differently.
whosgonnai would hope that delta would capture decrements, as i can also invision scenarios where that would be desirable.
whosgonnaso, maybe i monitor the actual value, and then trigger based on changes.
troffaskywhat type of data is this?
whosgonnawindows performance coutner for discarded packets
whosgonnasomeone sent me the current result of the counter saying that it's value is cumulative since reboot, and that it's current value of 131 is problematic (it's been six months since the host was rebooted). They can't even say when it occured, so ok - i can monitor it, and trigger when it changes.
whosgonnamaybe i should still look at it as a rate, though.
troffaskyah so not a normal throughput counter then
whosgonnano, it's not like bandwidth.
whosgonnabut - i suppose i could look at it that way.
whosgonnadropped packets _per_second_
troffaskyif you treat it as a throughput counter and trigger when the rate exceeds zero, then zabbix's putative rollover detection can handle it and your trigger can be simpler
troffaskyif you just trigger on change then you have to handle [or ignore] going back to zero
whosgonnathat shoudl be easy, because i can just subtract last value from second to last value, and trigger if the value is greater than zero.
whosgonnagoing from 200 to 0 would yield a value of -200, which is certainly not greater than zero :)
troffaskyyeah, that's what i meant about keeping you trigger simple :-)
whosgonnabiggest drawback is that it makes me have to manually calculate the loss when it occurs. That's not a huge deal of course, it's just not as clean.
whosgonnaor maybe i just still do the delta and ignore the resets if i have to graph it out, etc.
whosgonnatrigger is still on if the most recent value is greater than the previous value
SecHi. I have a question: A web scenario failed a step, but the "Last error message" is empty, and the Response code is marked as 0 for that moment. What would cause this?
Secah, nvm. found in the zabbix server log "Timeout was
Secreached: Operation timed out after 15001 milliseconds with 0 bytes received
heller_hey guys
heller_i need to use net.tcp.service.perf[http,,8080] but all i get is 0 as an output
whosgonnathat would indicate that the service is down according to the documentation.
whosgonnadoes `curl ` return a web page?
heller_ah, it gives a error 401
heller_seems like the http service which it should watch, needs an authentication
whosgonnaif you just want to see if there's ANY response, you can try telnet instead of http as the first argument
thclprHello everyone
zabbixHello thclpr, and welcome to #zabbix. I'm your friendly house bot :) Please be patient when it takes a while for people to reply. See our channel guidelines on and have a nice stay!
whosgonnabut, i think it sounds like you really want to do a webcheck
thclprI'm desperate and i have no idea where to ask help.... so i think this could a be a good place to finally get this issue regarding zabbix compilation and libevent dependency...
thclprany good soul available to help me on this matter?
whosgonnathclpr: on what platform?
thclprredhat 6
thclpr6.9 to be more exact
thclprbasically the problem is that i'm a server that i don't have root access
heller_whosgonna: well i need some sort of performance status
heller_to see how fast it will react
thclprI'm trying to compile zabbix-3.4.2 with the following params:
thclpr./configure --prefix=/storage/opt/zabbix-3.4.2 \ --enable-server \ --enable-java \ --enable-agent \ --with-postgresql \ --with-libpcre=/storage/opt/utils/pcre \ --with-libpcre-include=/storage/opt/utils/pcre/include/ \ --with-libevent=/storage/opt/utils/libevent \ --with-lxml2=/usr/bin/xml2-config \
whosgonnaheller: webchecks will do that- you can cover whole scenarios, INCLUDING the login that you're currently failing ;)
thclprlibevent was compiled by hand, only for zabbix usage
heller_oh how?
thclprversion is : libevent-2.1.8-stable
thclpr./configure runs well, but when i try to make install...
whosgonnaheller: big subject. Start here:
thclpr../../../include/sysinc.h:425:20: error: event.h: No such file or directory
whosgonnaheller, it will regurn response code, response time, and speed (throughput as a result of page size)
whosgonnaof course, you're getting a 401 on your request, so i'm guessing the page you're requesting is using http auth, so you could probably embed the credentials into your URL:
whosgonnaof course password in the URL isn't the best security practice ;)
heller_whosgonna: well actually i could even be happy just to see how fast will the 401 return
volterthclpr: Why do you want to run a Zabbix server on this host? Do you have good reasons?
thclprclient rquirement
volterAnd they won't install packages for you?
volterWho's going to maintain this later?
thclprlibevent-devel isn't available for that version of red hat enterprise
thclprme =(
volterCan't they provide you with a container?
thclprnot a option too
thclprmy only option is to use a custom installation of libevent
volterDo you want to run the frontend on this host too?
volterThe PHP version is too old then too.
thclprnot at all
thclpri compiled as well.
volterOh god
volterWell, your support matrix is growing.
thclprbut well.. any clue regarding how can i solve the libevent dependency? =)
volterI have to get off the train now, but we will find something.
thclpri tried the forums but it seems it won't accept new users
thclprand when i try to recover a possible user that i believe i created a few years back
thclprit asks 6 plus 2 for "robot check"
thclprand it says 8 is a wrong question ( =! )
thclpr( ?! )
heller_whosgonna: im going to use and set the login to the authentication page but it says connection refused
heller_whosgonna: im not sure if it does pass the login at all?
volterthclpr: Solved what?
thclprthe problem that i had regarding libevent
volterAny questions left open then?
thclprnot at all.
volterCongratulations then!
paulwasianyone have tips for troubleshooting high load on your zabbix server? I tried to disable an item check that is on a template linked to 900+ hosts, and the webpage just changes to a blank page without a spinning hourglass. URL: /zabbix/items.php?group_itemid[]=12345&hostid=123456&action=item.massdisable
paulwasiand the reason I mention load is we have an external monitoring program which reported high memory % utilization on the machine running zabbix
pbbunny0_happy holidays!!!
tuxickpaulwasi: add memory :)
paulwasiwhich holiday is today?
tuxick3 weeks before winter solstice
tuxickand crazy thursday
paulwasiI just want to learn how to watch/set/monitor it, because I know in the MSSQL and MySQL world, the DB engine will utilize 100% of the system memory if you don't explicitly restrict it
tuxickpaulwasi: yes, you need to tune mysql
paulwasiI'm not sure if Postgres which my zabbix server uses has the same behavior
tuxickoh postgres
tuxickwhatever :)
tuxickadd memory anyway
tuxicki bet you now have high iowait
paulwasialso, is there danger in doing service postgres restart or service zabbix-server restart when you think it's overloaded? will transactions be left in an inconsistent state?
paulwasi%Cpu(s): 10.3 us, 7.7 sy, 0.0 ni, 29.7 id, 51.6 wa, 0.0 hi, 0.7 si, 0.0 st
paulwasithat's 50% iowait, right?
paulwasiis that high?
tuxickquite :)
paulwasithat smiley is backward :(
tuxickso system is swapping or db is waiting for disk
paulwasiok, that sounds like I'm writing too many item checks to the DB too frequently, no?
tuxickiotop is rather useful
paulwasi(or mem paging would do it too, yes)
paulwasi29406 be/4 postgres 9.23 M/s 0.00 B/s 0.00 % 97.37 % postgres: zabbix zabbix ::1(53718) DELETE
paulwasilooks like garbage collector purging history is killign it
tuxick"as usual"
tuxickon zabbix server deleting history of 4 items would already kill server
tuxickzabbix server at work
tuxickalready demanded bigger hardware :)
tuxickmysql in our case
tuxicknot sure how in postgresql
tuxickbut in mysql you can kill a specific process
tuxickit's safe enough to kill that mass delete if you can
tuxickit's what i tend to do to recover live server
paulwasiyou need to do something about it before the next hourly purge runs though to prevent recurrence though, right?
tuxickwell it coincided with a change you made
tuxickgive that time to finish :)
paulwasiI actually tried to make that change to free up system resources. the memory alert came before I tried to disable the items
paulwasiluckily the item was an externalcheck so instead I chmod -x the script
paulwasiany quick tips to make psql queries from the CLI readable? the wordwrap is teeeeeeeerible
volterThat's a bit of a vague question.
whosgonnaany quick tips to make psql queries from the CLI readable? the wordwrap is VERY teeeeeeeerible
whosgonna(now it's more specific) :)
paulwasiadvance apologies for wall of text
paulwasizabbix=> select * from pg_stat_activity where state='active' order by xact_start asc; datid | datname | pid | usesysid | usename | application_name | client_addr | client_hostname | client_port | backend_start | xact_start | query_start | state_change | waiting | state | backend_xid | backend_xmin |
paulwasiok it filtered out line 2, you get the idea though
whosgonnapaulwasi: specifiy what columns you want/need in your select is a quick way to make that more readable
paulwasihard to tell what they mean this early on. I'm trying to find the query being run by a particular system PID, but the PID column doesn't match the PID from TOP
paulwasiahhhhhh I see it now
tuxickselect * from pg_stat_activity where pid = 29406?
tuxickor just kill 29406?
tuxickthat's what works on mysql anyway
paulwasithe PID in TOP is column TID, not the one over in the command column postgres: zabbix zabbix ::1(--->53718<---) which I thought it was
paulwasisorry not top, iotop
paulwasiI wanted to get pgsql to cancel it rather than have the os kill the process
paulwasinot sure it matters, but seemed cleaner
paulwasi following that, the cancel function didn't seem to do it, but the terminate one did
paulwasiand for others looking into this, this query seems useful and readable: select pid,usename,backend_start,waiting,state,query from pg_stat_activity where query like 'delete%' order by backend_start asc limit 5;
paulwasithank you tuxick for the less on linux iowait and iotop
timelessanyone here use auto registration?
timelessACTION is wondering if there's a way to do it securely
timelessespecially, i'd like to have autoregistration assign PSKs to each agent...
timelessand i'd like to make sure that random systems can't register w/ my zabbix
timelessI think I can kinda cheat by having a zabbix proxy inside a firewall be the thing that listens for autoregistration and have my agents register to it (w/ an action condition on proxy=that proxy)
timelessbut... can i easily unproxy the agent once i'm registered?
timelessand... what about the psk part? :)
timelessACTION reads
paulwasiyou can use the string in the .conf, but it's sent in clear text. there is a dropdown on hosts you can change from proxy to none after the fact, but not sure if there's a mass-way to do it
paulwasi<tuxick> you still there?
paulwasiregarding housekeeper long deletes, I checked the item it's been working on for 30+ minutes, and it only has 4010 rows in the history_uint table it's trying to delete from, and it's not making any progress according to select count(*) from that table where itemid matches that in the long-running query
paulwasiI see now, the itemid is changing in subsequent checks of that long running query, it's not one query, it's one process executing a bunch of similar queries in a row
timelesspaulwasi: fwiw, i've found that it's helpful to watch the "start time" for queries, since it will shift as they make "vague progress" through nearly identical queries
timelessi suspect i can have an action run an api script to change from proxy to non proxy
timelessbut, that probably won't do me many favors
timelessi wonder if it makes more sense to just write a script that uses the api and does all the work
timelessACTION ponders
NorNorHey all, I have a screen with the "Hosts Info" Resource however for some reason despite my hosts being either reachable via SNMP or not all hosts appear as "unknown" for their availability. Does anyone know where this availability status is taken from?
volterIs this a "host screen"?
volterIf so investigate the prototype.
NorNorHey volter, no it is just screen and the resource is a "hosts info" of the specified hostgroup.
paulwasiyeah if you're going to API to modify auto-registration, just handle registration with API too
paulwasi<NorNor> Unknown ZBX box or SNMP as well?
paulwasiif the SNMP interface IP is reachable for a particular host, that availability box lights up green in my experience. it does take a few minutes sometimes if you added a ton of hosts at once
NorNorpaulwasi: Just SNMP, the box does light green however for some reason the widget in the screen is still showing every host as "unknown" :( sad times.
paulwasihmm I'm not familiar with that widget. can you send a link to a screenshot or maybe the page URL so I can load it for my box?
NorNorpaulwasi: Not really good with IRC I hope you got that.
paulwasiNorNor, I saw it now, but that picture doesn't say unknown anywhere on it. it looks like a configuration options for style of the display