[collectd] Collectd and Otus Project
Good afternoon, I'm very interested about DISC systems and how to monitor them. I intend to use all those concepts to my researches in masters. I met Otus Project in Google Code. Reading some files of this project, I found something like Collected daemon. I got a bit confused: Did Otus Project generate Collected daemon? If not, what the difference between them? Is there a Final Version of Otus available for download? Thank you! == Geraldo A Sarmento Nt MSc Candidate on Computer Science at Federal University of Campina Grande - UFCG Distributed Systems Laboratory Paraíba - Brazil == ___ collectd mailing list collectd@verplant.org http://mailman.verplant.org/listinfo/collectd
[collectd] [PATCH] Plugin cpu - added CPU percent and summary usage
Added HZ and for all systems numcpu defaults to 1. Now jiffies multiplied to
100/HZ.
In sysctl capable systems HZ extracted from kern.clockrate stathz. In linux
it's sysconf(_SC_CLK_TCK); For other systems it's defaults to 100 for now.
Added config options
Plugin cpu
PerCpuStatistic true
ScaleTo100 false
/Plugin
Now summary CPU statistic always collected (it's more usefull than percore).
PerCpuStatistic - also collect data for each cpu
ScaleTo100 - scale summary cpu data to 100% (divide by numcpu).
---
src/cpu.c | 209 +
1 files changed, 155 insertions(+), 54 deletions(-)
diff --git a/src/cpu.c b/src/cpu.c
index 12071a2..6f533e6 100644
--- a/src/cpu.c
+++ b/src/cpu.c
@@ -113,7 +113,6 @@ static int cpu_temp_retry_max = 1;
/* #endif PROCESSOR_CPU_LOAD_INFO */
#elif defined(KERNEL_LINUX)
-/* no variables needed */
/* #endif KERNEL_LINUX */
#elif defined(HAVE_LIBKSTAT)
@@ -121,15 +120,14 @@ static int cpu_temp_retry_max = 1;
# define MAX_NUMCPU 256
extern kstat_ctl_t *kc;
static kstat_t *ksp[MAX_NUMCPU];
-static int numcpu;
+//static int numcpu;
/* #endif HAVE_LIBKSTAT */
#elif CAN_USE_SYSCTL
-static int numcpu;
/* #endif CAN_USE_SYSCTL */
#elif defined(HAVE_SYSCTLBYNAME)
-static int numcpu;
+//static int numcpu;
# ifdef HAVE_SYSCTL_KERN_CP_TIMES
static int maxcpu;
# endif /* HAVE_SYSCTL_KERN_CP_TIMES */
@@ -141,10 +139,48 @@ static int maxcpu;
#elif defined(HAVE_PERFSTAT)
static perfstat_cpu_t *perfcpu;
-static int numcpu;
+//static int numcpu;
static int pnumcpu;
#endif /* HAVE_PERFSTAT */
+static const char *config_keys[] =
+{
+ PerCpuStatistic,
+ ScaleTo100,
+ NULL
+};
+static int config_keys_num = 2;
+
+static int numcpu = 1;
+static bool per_cpu_statistic = false;
+static bool scale = true;
+static int hz=100;
+
+
+static int cpu_config (const char *key, const char *value)
+{
+ if (strcasecmp (key, PerCpuStatistic) == 0)
+ {
+ if (IS_TRUE (value))
+ per_cpu_statistic = true;
+else
+per_cpu_statistic = false;
+ }
+ else if (strcasecmp (key, ScaleTo100) == 0)
+ {
+ if (IS_TRUE (value))
+scale = true;
+else
+scale = false;
+ }
+ else
+ {
+ return (-1);
+ }
+
+ return (0);
+}
+
static int init (void)
{
#if PROCESSOR_CPU_LOAD_INFO || PROCESSOR_TEMPERATURE
@@ -199,6 +235,21 @@ static int init (void)
sstrerror (errno, errbuf, sizeof (errbuf)));
return (-1);
}
+
+struct clockinfo ci;
+ size_t ci_size = sizeof (ci);
+ int mib[2] = {CTL_KERN, KERN_CLOCKRATE};
+
+ status = sysctl (mib, STATIC_ARRAY_SIZE (mib),
+ ci, ci_size, NULL, 0);
+ if (status == -1)
+ {
+ char errbuf[1024];
+ WARNING (cpu plugin: sysctl: %s,
+ sstrerror (errno, errbuf, sizeof (errbuf)));
+ return (-1);
+ }
+hz = ci.stathz;
/* #endif CAN_USE_SYSCTL */
#elif defined (HAVE_SYSCTLBYNAME)
@@ -214,6 +265,17 @@ static int init (void)
return (-1);
}
+struct clockinfo ci;
+ size_t ci_size = sizeof (ci);
+ if (sysctlbyname (kern.clockrate, ci, ci_size, NULL, 0) 0)
+ {
+ char errbuf[1024];
+ WARNING (cpu plugin: sysctlbyname(kern.clockrate): %s,
+ sstrerror (errno, errbuf, sizeof (errbuf)));
+ return (-1);
+ }
+hz = ci.stathz;
+
#ifdef HAVE_SYSCTL_KERN_CP_TIMES
numcpu_size = sizeof (maxcpu);
@@ -238,6 +300,11 @@ static int init (void)
/* nothing to initialize */
#endif /* HAVE_PERFSTAT */
+#if defined(KERNEL_LINUX)
+numcpu = sysconf( _SC_NPROCESSORS_ONLN );
+hz = sysconf(_SC_CLK_TCK);
+#endif /* KERNEL_LINUX */
+
return (0);
} /* int init */
@@ -246,13 +313,20 @@ static void submit (int cpu_num, const char
*type_instance, derive_t value)
value_t values[1];
value_list_t vl = VALUE_LIST_INIT;
- values[0].derive = value;
+if (hz != 100)
+ values[0].derive = (derive_t) (value * (100.0 / hz));
+else
+ values[0].derive = value;
+
+if (scale cpu_num 0)
+values[0].derive = values[0].derive / numcpu;
vl.values = values;
vl.values_len = 1;
sstrncpy (vl.host, hostname_g, sizeof (vl.host));
sstrncpy (vl.plugin, cpu, sizeof (vl.plugin));
- ssnprintf (vl.plugin_instance, sizeof (vl.plugin_instance),
+if (cpu_num = 0)
+ ssnprintf (vl.plugin_instance, sizeof (vl.plugin_instance),
%i, cpu_num);
sstrncpy (vl.type, cpu, sizeof (vl.type));
sstrncpy
Re: [collectd] Simple Tail plugin regex -- count lines
Martin, The part I struggled with at first is the data is going to be converted into a rate - so the number represents Transactions per second. I found on my use of tail I needed to increase the polling interval so that enough transactions were caught between polls to actually create meaningful data points. In the global section I went to: Interval 300 10,000 records in an hour would still only be 3 tps In my tail match section using: DSType CounterInc Type invocations Hope it helps Bob -Original Message- From: collectd-boun...@verplant.org [mailto:collectd-boun...@verplant.org] On Behalf Of Martin B. Smith Sent: Monday, March 14, 2011 11:54 AM To: collectd@verplant.org Subject: [collectd] Simple Tail plugin regex -- count lines Hi all, I've been struggling with the regex and/or dstype of the tail plugin. I'm basically wanting to count lines in an audit file that logs authentications. Any line should match. Here's what I've got: LoadPlugin tail Plugin tail File /var/log/shibboleth-idp/idp-audit.log Instance idp_audit_log Match Regex .* DSType CounterInc Type derive Instance urn_ufl_edu /Match /File /Plugin I know that the log has more lines in it, since I can run this command and see the line count jump dramatically even in just a few seconds: $ cat /var/log/shibboleth-idp/idp-audit.log | wc -l Unfortunately, when I dump the rrd, I get AVERAGE values like: !-- 2011-03-14 11:15:00 EDT / 1300115700 -- rowv 1.915600e+00 /v/row !-- 2011-03-14 11:20:00 EDT / 1300116000 -- rowv 1.963867e+00 /v/row !-- 2011-03-14 11:25:00 EDT / 1300116300 -- rowv 2.103644e+00 /v/row !-- 2011-03-14 11:30:00 EDT / 1300116600 -- rowv 1.989200e+00 /v/row !-- 2011-03-14 11:35:00 EDT / 1300116900 -- rowv 1.953467e+00 /v/row I *know* I'm seeing more than 1-2 lines in a 5 minute interval. Is it possible the numbers are too large that I'm substracting, yielding an overflow? A single hour can sometimes have 10k lines in the log file. Thanks in advance for your consideration! -- Martin B. Smith smit...@ufl.edu - (352) 273-1374 CNS/Open Systems Group University of Florida ___ collectd mailing list collectd@verplant.org http://mailman.verplant.org/listinfo/collectd
Re: [collectd] NetApp plugin giving NANs
On 15.03.2011 12:40, Mark Syms wrote: Hi, I've been trying to use the NetApp plugin to monitor a NetApp Fas2020 that I'm working with. Collectd is running on Debian, I've built a modified package of version 4.10.1-1 using the NetApp SDK to allow enabling the NetApp plugin. The NetApp is running version 7.2.6.1 of the system software. rrdtool info for the cpu-system.rrd file gives the following (the other RRDs are similar). what interval did you configure for the NetApp plugin? There used to be a bug that caused this kind of behavior if the global interval was different from the interval used by the netapp plugin. If this error condition fits your setup, you might want to update to version 4.10.2. After this you have to delete all empty rrd files of the netapp plugin, as they have been created with with the wrong interval setting. If updating is not an option for whatever reason you can adjust the interval of the netapp plugin to your global interval, which should fix the problem immediately. Or, if you know a lot more about rrd than I do, you could edit the already created rrd files and adjust their update interval, but I have no idea how to do that. But someone else on this mailing list might be able to answer this question. Regards, Sven ___ collectd mailing list collectd@verplant.org http://mailman.verplant.org/listinfo/collectd
Re: [collectd] [Bug 734179] collectd runs as root unnecessarily
Hi Andrew and Sebastian, On Sun, Mar 13, 2011 at 02:39:06PM +0100, Sebastian Harl wrote: With this E-mail, I'm also submitting this as wishlist bug report to the Debian BTS and I'm going to fix it in my next upload. I would have liked to answer to that bug as well, but couldn't find it on bugs.debian.org. Am I missing something? On Sun, Mar 13, 2011 at 07:04:39AM -, Andrew Yates wrote: (It might also be a good idea to run collectd as an unprivileged user by default. A quick test suggests that of the default plugins only 'df' may require root privileges.) In the vast majority of cases, the df plugin doesn't need privileges. The only problematic case I can think of right now is if some file system is mounted at /secret/foo and the unprivileged user isn't allowed to read /secret. Plugins which do require privileges can be found in the collectd wiki in the Plugins requiring privileges category [0]. Regards, —octo [0] http://collectd.org/wiki/index.php/Category:Plugins_requiring_privileges -- Florian octo Forster Hacker in training GnuPG: 0x0C705A15 http://octo.it/ signature.asc Description: Digital signature ___ collectd mailing list collectd@verplant.org http://mailman.verplant.org/listinfo/collectd
