05/28/2009 02:00:34 PM - SPINE: Poller[0] Time: 31.9659 s, Threads: 10, Hosts: 2265
05/28/2009 02:00:34 PM - SYSTEM STATS: Time:32.6916 Method:spine Processes:1 Threads:10 Hosts:2265 HostsPerProcess:2265 DataSources:14022 RRDsProcessed:9351
05/28/2009 02:05:02 PM - POLLER: Poller[0] WARNING: Poller Output Table not Empty. Issues Found: 1, Data Sources: (DS[9532])
mysql> select * from poller_output;
+---------------+----------+---------------------+-------------+
| local_data_id | rrd_name | time | output |
+---------------+----------+---------------------+-------------+
| 9532 | | 2009-05-28 14:00:05 | rta:15.9821 |
+---------------+----------+---------------------+-------------+
For the record, it always seems to be the same hosts too.
I believe I've caught all the logs at the various levels. I also have a some what convoluted way of getting data into cacti, hopefully that's not where the problem lies!
I'm using the script server to fetch the data. The 'ss_fetch_nagios_perf_web' retrieves the nagios performance data from a database through a simple web service.
05/28/2009 02:00:05 PM - SPINE: Poller[0] Host[37] DS[507] SS[0] SERVER: /var/www/cacti/scripts/ss_fetch_nagios_perf_web.php ss_fetch_nagios_perf_web ns1 ping, output: rta:3.3261
05/28/2009 02:00:05 PM - SPINE: Poller[0] Host[37] DS[9532] SS[1] SERVER: /var/www/cacti/scripts/ss_fetch_nagios_perf_web.php ss_fetch_nagios_perf_web ns1 dns_stats, output: rta:15.9821
The ping output is correct, however the dns_stats output is wrong. It's actually ping output as well!
I can see the incoming web service requests in the apache logs. The returned data size is spot on for the two different metrics.
127.0.0.1 - - [28/May/2009:14:00:05 +1000] "GET /perl/nagios-perf?host=ns1&stat=ping HTTP/1.1" 200 9 "-" "-"
127.0.0.1 - - [28/May/2009:14:00:05 +1000] "GET /perl/nagios-perf?host=ns1&stat=dns_stats HTTP/1.1" 200 61 "-" "-"
I also enabled the mysql logging and everything is in order there too. It looks like the script server is getting confused handling all the data it gets back??
The data cacti logged for the dns_stats is actually ping data from a totally different host!
| 470891 | 2009-05-28 13:59:23 | redback | ping | 0.099 | 0.595 | rta=15.982ms;500.000;4000.000;0; pl=0%;80;100;; |
I'll switch back to running my perl script to fetch all this data to see if it improves. However I was enjoying the improved performance when using the script server. I don't think it's my dodgy php script..... When I test it via the php script server it works as expected for both ping and dns_stats. This issue only happens intermittently as well which makes it odd.
For the record, running cacti 0.8.7d and spine 0.8.7c.
Any thoughts welcomed!
Regards,
ajt.
