snmp output U

bsellis · Post by **bsellis** » Wed Feb 02, 2005 5:16 pm

Starting last week, Cacti is returning bad snmp data. The rrd files are being updated, but the graph is showing no data (but the times are being updated). THe current/max/avg are showing 'nan' next to them.

This only is happening for certain OID's such as partition info and CPU usage on Windows servers. All data for my NetWare servers are being graphed fine.

Upgraded from Cacti 0.8.6b to 0.8.6c today. Also cleared the poller cache and truncated the poller output table (as suggested in other postings..I checked first!) I am using cmd.php, not cactid

Here is an example from the log file. Note that I substitued in for IP address and snmpstring

02/02/2005 04:43:20 PM - CMDPHP: Poller[0] Host[59] SERVER: /Users/cactiuser/scripts/ss_host_cpu.php ss_host_cpu <!ipaddresshere> <snmpstringhere> 1 get usage 1, output: U
02/02/2005 04:43:20 PM - CMDPHP: Poller[0] Host[59] WARNING: Result from SERVER not valid. Partial Result:

Any help would be appreciated! Thanks!

Post by **TheWitness** » Wed Feb 02, 2005 7:50 pm

If you have net-snmp (www.net-snmp.org) installed on your server, you can run "snmpwalk -c <mycommunity> -v <mysnmpversion> <myhostname>" and it should return data. If not, you may have a snmp problem. Try that first.

TheWitness

bsellis · Post by **bsellis** » Wed Feb 02, 2005 8:42 pm

I ran an snmpwalk...and a ton of data returned. I also ran a Verbose Query and it walked fine. I forgot to mention that SOME of the graphs on these devices are working. Although CPU, partition information, and virutal memory are missing, the Processes and Network Interface graphs are working fine. But three other servers have this same problem--only showing these two graphs. Thanks

Post by **TheWitness** » Wed Feb 02, 2005 8:48 pm

Ok,

OS?
PHP Version?
PHP CLI or CGI?

Then, review your poller cache for the statement "Action 2:". In that line, you will find something along the line of:

<scrpt_path>/<script_name> <scrpt_name> <arguments>

Copy that information to the clipboard. Then manually start the script server. To do that open a command line and type the following:

php <cactipath>/scrip_server.php

You will get a line saying something like "Script Server Started..."

Now paste the information you copied from the previous section into the command line, hit enter and data should be returned.

If you get any warning or error messages when either starting script server or in your response, this will cause issue with the script server.

To exit the script server type "quit"

TheWitness

bsellis · Post by **bsellis** » Wed Feb 02, 2005 9:13 pm

Hello:

OS: Mac OS X 10.3.7
PHP v4.3.10
PHP CLI

Code: Select all

Action: 2, Script: /Users/cactiuser/scripts/ss_host_cpu.php ss_host_cpu <ipaddress> <snmpstring> 1 get usage 1

Code: Select all

/Users/cactiuser/scripts/ss_host_cpu.php ss_host_cpu <ipaddress> <snmpstring> 1 get usage 1
80

That result makes sense..80% CPU. I ran it another time and got 94. One other thing I should mention is that last week, we had a power outage, and no rrd files were being updated at all (everything had been working fine before that for 3 months). At the advice of another posting, we truncated the poller_output table. This fixed all of the netware servers and some graphs on the windows servers.

ppollet73 · Post by **ppollet73** » Thu Feb 03, 2005 2:43 am

Did you change your webserver root ?
I had the same problem which disappear when I move back all the cacti to the default website root (in my case with linux /var/www/html)

HTH

Pipol

Post by **TheWitness** » Thu Feb 03, 2005 5:47 am

You may have corrupted RRD files. You should delete one (keep a backup) and see it it starts updating again. You can also run poller.php in debug from the command line to view the RRD stdout/stderr.

There has been a history of system crashes causing corrupt RRD files.

TheWitness

bsellis · Post by **bsellis** » Thu Feb 03, 2005 1:46 pm

I am using the default web root path. A symbolic link points to the cactiuser home directory from web root. I deleted one of the rrd files. The graph recreated with no data.

When running poller.php in debug:

there were a lot of OK's, but here are the errors:

Code: Select all

ERROR: illegal attempt to update using time 1107454730 when last update time is 1107454730 (minimum one second step)
OK u:0.05 s:0.10 r:128.60

Code: Select all

Error in packet
Reason: (noSuchName) There is no such variable name in this MIB.
Failed object: UCD-SNMP-MIB::laLoad.1

Code: Select all

Timeout: No Response from <ipdaddress>:161.
Error in packet
Reason: (noSuchName) There is no such variable name in this MIB.
Failed object: SNMPv2-SMI::enterprises.9.2.1.58.0

Not sure why SNMPv2 is there ---it is not enabled on my server and is snmp version 1 is chosen in the device properties for that server.

Post by **TheWitness** » Fri Feb 04, 2005 8:10 pm

Hello:

Code:
ERROR: illegal attempt to update using time 1107454730 when last update time is 1107454730 (minimum one second step)
OK u:0.05 s:0.10 r:128.60

This error is typically indicative of 2 cron jobs running simultaneously

Code:
Error in packet
Reason: (noSuchName) There is no such variable name in this MIB.
Failed object: UCD-SNMP-MIB::laLoad.1

This error means that the OID in question does not exist for you SNMP agent.

Code:
Timeout: No Response from <ipdaddress>:161.
Error in packet
Reason: (noSuchName) There is no such variable name in this MIB.
Failed object: SNMPv2-SMI::enterprises.9.2.1.58.0

IBID from above.

TheWitness

bsellis · Post by **bsellis** » Mon Feb 07, 2005 11:26 am

Hello,

The OIDs that I am having problems with are valid. When I ran the poller in debug it returned correct values (such as 80 for CPU usage). I also tested using Getif. But in here they are returning a U (see below) and displaying 'nan' in the graph, but the graphs are being updated each polling cycle with no data. This is happening on all of my Windows servers except one, and is happening on a couple of NetWare servers. Logged in users, processes, and interface info are being graphed fine, but not CPU, Disk/volume info, or memory. They stopped working after a power outage which only affected the cacti server and not any other server. At that point, none of the rrd files were being updated, so we truncated the poller_output table, which resolved all graphing problems (for the most part) for my NetWare servers, but the Windows servers are still not graphing properly. Additionally, when I add new Windows servers, I experience the same problem.

Code: Select all

02/02/2005 04:38:13 PM - CMDPHP: Poller[0] Host[59] SERVER: /Users/cactiuser/scripts/ss_host_disk.php ss_host_disk <ipaddress> <snmpstring> 1 59 get total 7, output: U
02/02/2005 04:38:13 PM - CMDPHP: Poller[0] Host[59] WARNING: Result from SERVER not valid.  Partial Result: 
02/02/2005 04:38:13 PM - CMDPHP: Poller[0] Host[59] SERVER: /Users/cactiuser/scripts/ss_host_disk.php ss_host_disk <ipaddress> <snmpstring> 1 59 get used 7, output: U
02/02/2005 04:38:13 PM - CMDPHP: Poller[0] Host[59] WARNING: Result from SERVER not valid.  Partial Result: 
02/02/2005 04:38:13 PM - CMDPHP: Poller[0] Host[59] SERVER: /Users/cactiuser/scripts/ss_host_cpu.php ss_host_cpu <ipaddress> <snmpstring> 1 get usage 0, output: U
02/02/2005 04:38:13 PM - CMDPHP: Poller[0] Host[59] WARNING: Result from SERVER not valid.  Partial Result: 
02/02/2005 04:38:13 PM - CMDPHP: Poller[0] Host[59] SERVER: /Users/cactiuser/scripts/ss_host_cpu.php ss_host_cpu <ipaddress> <snmpstring> 1 get usage 1, output: U

Post by **TheWitness** » Mon Feb 07, 2005 5:29 pm

You should force a reindex by clicking on the little green circle to the right of the data queries when you edit the host.

TheWitness

bsellis · Post by **bsellis** » Mon Feb 07, 2005 6:49 pm

Thanks - I just tried that on a couple of them but it didn't seem to help. I also added another Windows server, and the same graphs are not showing data. Do you think it is a host template issue? If so, how would I check that? The host template is the only other common denominator that I can think of that would be causing the problem.

Thanks

Post by **TheWitness** » Tue Feb 08, 2005 5:53 am

To debug the script server, you need to perform the following:

1) Start the script server in debug mode: php /<webroot>/script_server.php

If you get any warnings other than "Script Server Started..." then there is a problem and it should be corrected.

2) Run your script. The syntax is the same as your post:

/Users/cactiuser/scripts/ss_host_cpu.php ss_host_cpu <ipaddress> <snmpstring> 1 get usage 0

You should receive a response.

3) Quit the script_server by typing "quit"

Please post your results.

TheWitness

bsellis · Post by **bsellis** » Wed Feb 09, 2005 9:54 am

Hello again:

Code: Select all

/Users/cactiuser/scripts/ss_host_cpu.php ss_host_cpu <ipaddress> <snmpstring> 1 get usage 0
83

Post by **TheWitness** » Wed Feb 09, 2005 5:28 pm

Did you receive any warnings when starting up the script server? Did you receive any errors when clearing your poller cache?

Please make a couple of shot's and post them:
1) Show Broken Graph
2) Show Broken Graph Data Sources in EDIT Mode
3) Show Host Edit Screen for Broken Graphs.

Thanks,

TheWitness

snmp output U

snmp output U

Who is online