ERROR: opening '/...rrd' : No such file or directory

[mdurand]

Hello all,

I've got a problem with the rrd files; and i found no response on this forum.

I made my own template.
When i created only one graph: the rrd file was created immediately.
But when i created several graphs (about 30 with a script that i wrote) Cacti took 3 hours to create the rrd files.

Now with my script i want to create about 100 graphs but i waited for 20hours and no rrd file has been created.

I would like to know if i have to do something to make the rrd files been created. And, if not, why does it take so much time?

If you don't know how to resolve my problem tell me and i will know how much difficult it is to solve my problem. If you want more information you can ask.

ps: please excuse me if my english is a little bad (i'm french)

[gandalf]

cacti creates rrd files with the next polling cycle, assuming the target "host availabilty method" sees this host as "up" (and file permissions ... are ok). Please increment php.ini's memory size to 64 MB (default is 8MB), this may help as well
Reinhard

[mdurand]

cacti creates rrd files with the next polling cycle, assuming the target "host availabilty method" sees this host as "up" (and file permissions ... are ok). Please increment php.ini's memory size to 64 MB (default is 8MB), this may help as well
Reinhard

i already changed 'memory size' in php.ini
but 'host availability' is to 'unknown' instead of 'up'
i think it's because cacti doesn't try to ask something to the host anymore
how can i change it?

[gandalf]

Bonjour
You're quite right in assuming that cacti won't create anything for hosts in "unknown" state. Unless you request any graphs for a new host, cacti's poller won't be aware of it. And then, it will cange "unknown" to down", if the host i snot reachable (depending on "Downed Host Detection") or to "up" if the host is reachable.
You may want to change "Settings -> Poller Logging Level" to DEBUG for one polling cycle to get more info
bonne chance
Reinhard

[mdurand]

hi,

i am in debug mode but i don't know what to look for

when i look at my devices: although if they are 'unknown' cacti knows somethings that i don't precised
for instance: snmp informations

so i think that cacti manage to get informations
but it don't try to get data for my graphs

thanks

[xorian]

Well, I have the same problem. The .RRD files are simply not created, and the status is "Unknown". Guess that's the problem, but any idea how to fix it? I've already gone over all the possible solutions.

[mdurand]

today my graphs were created by cacti which get data from the host
but i didn't do anything for that or i don't know what i did

i don't understand what happened
i created the graphs 1week ago

today i just tried to create a new graph
this graph hasn't been created by cacti yet
but all the others are

please help me to understand

[mdurand]

note for the previous post:

my devices still are 'unkown' but the graph are updated since half an hour

[mdurand]

i would like to precise that :

to add my graphs in cacti i manipled the database
i didn't used the cacti interface

so it is possible that in the interface there's a submit button which make cacti create the graph

[gandalf]

I've described above the procedure cacti uses to create rrd files. To be more specific: The web interface places some "rrdtool create" statements into mysql which will be processed by the poller on the next run. So you'll have to make sure the poller is running fine, else nogo.
The poller first checks each host to be executed for "Downed Host Detection". If the poller detects, that the Host is not reachable. it will skip ALL data gathering and rrd file creation for THAT VERY host.
Host detection can be skipped for a specific host, when omitting snmp community string in the host's definition within cacti.
Else, you'll have the option to check for PING (ICMP, UDP, TCP) reachability and/or SNMP reachability. If those detection methods fail, cacti's poller will again skip all further actions related to that very host.
As soon as the poller ONCE detects the hosts as UP, it will create the needed rrd files. If it detects the hosts as down later, it will again skip data gathering (but the rrd files may be still present).

So it's cruical, that poller's Host Detection goes well. You may see this in cacti.log, when running DEBUG mode (Settings -> Logging Level).
Failing check goes like

Code: Select all

07/25/2006 06:20:17 PM - CACTID: Poller[0] Host[10] SNMP Result: Host did not respond to SNMP
07/25/2006 06:20:17 PM - CACTID: Poller[0] Host[10] DEBUG: HOST COMPLETE: About to Exit Host Polling Thread Function

Omiiting the availability check will result in (using Advanced Ping Template)

Code: Select all

07/25/2006 06:25:06 PM - CACTID: Poller[0] Host[5] SNMP Result: Device does not require SNMP
07/25/2006 06:25:06 PM - CACTID: Poller[0] Host[5] No host availability check possible for 'www.lvm.de'
07/25/2006 06:25:07 PM - CACTID: Poller[0] Host[5] DS[51] SS[1] SERVER: /var/www/html/workspace/branch/scripts/ss_fping.php ss_fping www.lvm.de 20 TCP 80, output: min:26.0500 avg:37.9690 max:64.8800 dev:6.8905 loss:0.0000

Normal check (successful)

Code: Select all

07/25/2006 06:20:05 PM - CACTID: Poller[0] Host[2] SNMP Result: Host responded to SNMP
07/25/2006 06:20:05 PM - CACTID: Poller[0] Host[2] DS[58] SNMP: v1: gandalf, dsname: uptime, oid: .1.3.6.1.2.1.1.3.0, value:
343733
...

hth
Reinhard

[mdurand]

thanks for your reply

these are my news:

my rrd files have been created and i receive data
my devices were unknown but it worked in spite of it

in my cacti.log file i can read (at the date when i first received data)

07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: Device does not require SNMP
07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: v2: 10.234.192.5, dsname: total_mem, oid: .1.3.6.14.11.5.1.1.2.1.1.1.5.1, output: 7940176

it begin like you said when host is unknown but then i received data (i created the rrd files handly before)

note: today i added to my devices some data query and their status became up.

My problem today is that i don't know why the poller took so much time to check the device (1 week)

maybe like you said the poller doesn't run well
but i have some others graphs which are regularly updated, so i think that the poller run correctly
am i right?

[gandalf]

thanks for your reply

these are my news:

my rrd files have been created and i receive data
my devices were unknown but it worked in spite of it

in my cacti.log file i can read (at the date when i first received data)

07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: Device does not require SNMP
07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: v2: 10.234.192.5, dsname: total_mem, oid: .1.3.6.14.11.5.1.1.2.1.1.1.5.1, output: 7940176

Well, that's the "case 1" from above. This one holds, when you won't need snmp access to the target device. In this case, you'll leave the community string empty (as it is not needed) to inform cacti to omit host availability checks.
But as soon as you're going to request data via SNMP, this will not be a good idea anymore. You'll need (AFAIK) the snmp community string to authenticate against the targets snmp daemon. So you'll have to enter it as cacti will start to check host availabilty (I'm not sure whether this one needs a "Clear Poller Cache" which in fact rebuilds the poller cache)

it begin like you said when host is unknown but then i received data (i created the rrd files handly before)

note: today i added to my devices some data query and their status became up.

My problem today is that i don't know why the poller took so much time to check the device (1 week)

maybe like you said the poller doesn't run well
but i have some others graphs which are regularly updated, so i think that the poller run correctly
am i right?

Status "unknown" is the initial status for all devices freshly added to cacti. This will stay forever, until some graph is requested for the host (host availabilty checks if not a single graph is requested would be rather unnecessary). As soon as you request at least one graph, the poller should perform host availabilty check (if community string given! See above!) starting with the next polling interval.
You may follow all this when leaving poller in DEBUG mode. But this is usually not a good recommendation unless you keep your eyes on the log's file size.

Reinhard

[mdurand]

i never let the community string empty
and when i get " CMDPHP: Poller[0] Host[4] SNMP: Device does not require SNMP " it wasn't empty

how do you explain that device doesn't require snmp
and the ligne just after it get data with snmp:
"07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: Device does not require SNMP
07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: v2: 10.234.192.5, dsname: total_mem, oid: .1.3.6.14.11.5.1.1.2.1.1.1.5.1, output: 7940176 "

status unknown:

This will stay forever, until some graph is requested for the host (host availabilty checks

no it has stayed even after getting data and drawing graph
the status changed when i add some data queries to the devices
and the community is given

thanks

[gandalf]

i never let the community string empty
and when i get " CMDPHP: Poller[0] Host[4] SNMP: Device does not require SNMP " it wasn't empty

how do you explain that device doesn't require snmp
and the ligne just after it get data with snmp:
"07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: Device does not require SNMP
07/24/2006 12:05:36 PM - CMDPHP: Poller[0] Host[4] SNMP: v2: 10.234.192.5, dsname: total_mem, oid: .1.3.6.14.11.5.1.1.2.1.1.1.5.1, output: 7940176 "

I assumed, that the target snmpd may substitue "public" as a default community string, when none given. It was quite astonishing for me, as the "SNMP: Device does not require SNMP" is known only for those cases I described.

status unknown:

This will stay forever, until some graph is requested for the host (host availabilty checks

no it has stayed even after getting data and drawing graph
the status changed when i add some data queries to the devices
and the community is given

thanks

Both cases are strange to me, will ask a dev
Reinhard

[mdurand]

lvm:
thanks a lot for the time and care you take for my problem

marion[/list]