Page tree
Skip to end of metadata
Go to start of metadata

 


Statistics

Statistics in collectd consist of a value list. A value list includes:

 

Value listExamplecomment
Values 99.8999percentage
Value lengththe number of values in the data set.  
Timetimestamp at which the value was collected.1475837857epoch
Intervalinterval at which to expect a new value.10interval
Hostused to identify the host.localhostcan be uuid for vm or host… or can give host a name
Pluginused to identify the plugin.cpu 
Plugin instance (optional)used to group a set of values together. For e.g. values belonging to a DPDK interface.0 
Typeunit used to measure a value. In other words used to refer to a data set.percent 
Type instance (optional)used to distinguish between values that have an identical type.user 
meta dataan opaque data structure that enables the passing of additional information about a value list. “Meta data in the global cache can be used to store arbitrary information about an identifier”   

Notifications


Notifications in collectd are generic messages containing:

An associated severity, which can be one of OKAY, WARNING, and FAILURE.
A time.      
A Message     
A host.      
A plugin.      
A plugin instance (optional).    
A type.      
A types instance (optional).    
Meta-data.     


Example notification:

 

Severity:FAILURE
Time:1472552207.385
Host:pod3-node1
Plugin:dpdkevents
PluginInstance:dpdk0
Type:gauge
TypeInstance:link_status
DataSource:value
CurrentValue:1.000000e+00
WarningMin:nan
WarningMax:nan
FailureMin:2.000000e+00
FailureMax:nan
Hostpod3-node1, plugin dpdkevents (instance dpdk0) type gauge (instance link_status): Data source "value" is currently 1.000000. That is below the failure threshold of 2.000000.

 

Supported Metrics and Events

Recommended Intervals for plugins

  • For events plugins: In order to meet a 10ms detection time, it's recommended to use a 5ms interval for the events plugins.
  • For stats plugins: generally these can run at the global interval configured for collectd. Where there is an exception this is pointed out in the table below.

Metrics

Reference starting point: https://github.com/collectd/collectd/blob/master/src/types.db  

But below is a mapping of the "base" plugins that would run on the host/the guest.

 

 

 

Where collectd is runningPluginPlugin InstanceTypeType InstanceDescriptionRangecommentAdditional Info
Host/guest

CPU

(A read plugin that retrieves CPU usage in Nanoseconds of as a percentage)

 percent/nanosecondsidleTime CPU spends idle.  Can be per cpu/aggregate across all the cpus.For more info, please see:http://man7.org/linux/man-pages/man1/top.1.html
http://blog.scoutapp.com/articles/2015/02/24/understanding-linuxs-cpu-stats
Note that jiffies operate on a variable time base, HZ. The default value of HZ should be used (100), yielding a jiffy value of 0.01 seconds) [time(7)]. Also, the actual number of jiffies in each second is subject to system factors, such as use of virtualization. Thus, the percent calculation based on jiffies will nominally sum to 100% plus or minus error.
 
      
 percent/nanosecondsniceTime the CPU spent running user space processes that have been niced. The priority level a user space process can be tweaked by adjusting its niceness.  
 percent/nanosecondsinterruptTime the CPU has spent servicing interrupts.  
 percent/nanosecondssoftirq(apparently) Time spent handling interrupts that are synthesized, and almost as important as Hardware interrupts (above). "In current kernels there are ten softirq vectors defined; two for tasklet processing, two for networking, two for the block layer, two for timers, and one each for the scheduler and read-copy-update processing. The kernel maintains a per-CPU bitmask indicating which softirqs need processing at any given time." [Ref]  
 percent/nanosecondsstealCPU steal is a measure of the fraction of time that a machine is in a state of “involuntary wait.”  It is time for which the kernel cannot otherwise account in one of the traditional classifications like user, system, or idle.  It is time that went missing, from the perspective of the kernel.http://www.stackdriver.com/understanding-cpu-steal-experiment/  
    
 percent/nanosecondssystemTime that the CPU spent running the kernel.   
 percent/nanosecondsuserTime CPU spends running un-niced user space processes.    
 percent/nanosecondswaitThe time the CPU spends idle while waiting for an I/O operation to complete   

Interface

(A read plugin that retrieves Linux Interface statistics)

 if_droppedinThe total number of received dropped packets.   
 if_errorsinThe total number of received error packets. http://www.onlamp.com/pub/a/linux/2000/11/16/LinuxAdmin.html 
 if_octetsinThe total number of received bytes.   
 if_packetsinThe total number of received packets.   
 if_droppedoutThe total number of transmit packets dropped   
 if_errorsoutThe total number of transmit error packets. (This is the total of error conditions encountered when attempting to transmit a packet. The code here explains the possibilities, but this code is no longer present in /net/core/dev.c  master at present - it appears to have moved to /net/core/net-procfs.c.)   
 if_octetsoutThe total number of bytes transmitted   
 if_packetsoutThe total number of transmitted packets   

Memory

(A read plugin that retrieves memory usage statistics)

 memorybufferedThe amount, in kibibytes, of temporary storage for raw disk blocks. https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Deployment_Guide/s2-proc-meminfo.html 
 memorycachedThe amount of physical RAM, in kibibytes, left unused by the system.   
 memoryfreeThe amount of physical RAM, in kibibytes, left unused by the system.   
 memoryslab_reclThe part of Slab that can be reclaimed, such as caches. Slab — The total amount of memory, in kibibytes, used by the kernel to cache data structures for its own use 
 memoryslab_unreclThe part of Slab that cannot be reclaimed even when lacking memory https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Deployment_Guide/s2-proc-meminfo.html 
 memorytotalTotal amount of usable RAM, in kibibytes, which is physical RAM minus a number of reserved bits and the kernel binary code. https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Deployment_Guide/s2-proc-meminfo.htmlThis was the only undefined metric in the mem_used calculation below.
 memoryusedmem_used = mem_total - (mem_free + mem_buffered + mem_cached + mem_slab_total); https://github.com/collectd/collectd/blob/master/src/memory.c#L325 

disk

(A read plugin that retrieves disk usage statistics)

 disk_io_timeio_timetime spent doing I/Os (ms). You can treat this metric as a device load percentage (Value of 1 sec time spent matches 100% of load).   
 disk_io_timeweighted_io_timemeasure of both I/O completion time and the backlog that may be accumulating.   
 disk_mergedreadthe number of operations, that could be merged into other, already queued operations, i. e. one physical disk access served two or more logical operations. Of course, the higher that number, the better.   
 disk_mergedwritethe number of operations, that could be merged into other, already queued operations, i. e. one physical disk access served two or more logical operations. Of course, the higher that number, the better.   
 disk_octectsreadthe number of octets read from a disk or partition   
 disk_octectswritethe number of octets written to a disk or partition   
 disk_opsreadthe number of read operations issued to the disk   
 disk_opswritethe number of write operations issued to the disk   
 disk_timereadthe average time an I/O-operation took to complete. Note from collectd Since this is a little messy to calculate take the actual values with a grain of salt.   
 disk_timewritethe average time an I/O-operation took to complete. Note from collectd Since this is a little messy to calculate take the actual values with a grain of salt. https://collectd.org/wiki/index.php/Plugin:Disk 
 pending_operations shows queue size of pending I/O operations. http://lxr.free-electrons.com/source/include/uapi/linux/if_link.h#L43 

Ping

(A read plugin that retrieves the RTT for a ping)

 ping Network latency is measured as a round-trip time in milliseconds. An ICMP “echo request” is sent to a host and the time needed for its echo-reply to arrive is measured. Latency 
 ping_droprate droprate = ((double) (pkg_sent - pkg_recv)) / ((double) pkg_sent); https://github.com/collectd/collectd/blob/master/src/ping.c#L703 
 ping_stddev


 

if pkg_recv > 1

latency_stddev = sqrt (((((double) pkg_recv) * latency_squared) - (latency_total * latency_total)) / ((double) (pkg_recv * (pkg_recv - 1))));

 https://github.com/collectd/collectd/blob/master/src/ping.c#L698 
   pkg_recv = # of echo-reply messages receivedlatency_squared = latency * latency (for a received echo-reply message)latency_total = the total latency for received echo-reply messages 
    
    

load

(A read plugin that retrieves system load for 1, 5 and 15 mins.)

 load

shorttermload average figures giving the number of jobs in the run queue (state R) or waiting for disk I/O (state D) averaged over 1 Minute http://man7.org/linux/man-pages/man5/proc.5.html  
 measured CPU and IO utilization for 1 min using /proc/loadavg https://github.com/collectd/collectd/blob/master/src/load.c 
 midtermload average figures giving the number of jobs in the run queue (state R) or waiting for disk I/O (state D) averaged over 5 Minutes   
 measured CPU and IO utilization for 5 mins using /proc/loadavg   
 longtermload average figures giving the number of jobs in the run queue (state R) or waiting for disk I/O (state D) averaged over 15 Minutes   
 measured CPU and IO utilization for 15 mins using /proc/loadavg   

OVS events

(A read plugin that retrieves events (like link status changes) from OVS.)

 gaugelink_statusLink status of the OvS interface: UP or DOWN   

OVS Stats

(A read plugin that retrieves interface stats from OVS.)

   if_collisions Number of collisions. per interface 
   if_rx_octets Number of received bytes. http://openvswitch.org/ovs-vswitchd.conf.db.5.pdf 
   if_rx_errorscrcNumber of CRC errors.   
   if_dropped       rx: Number of packets dropped by RX.   
   if_errors           rx: Total number of receive errors, greater than or equal to the sum of the RX errors above.   
   if_rx_errorsframeNumber of frame alignment errors.   
   if_rx_errorsoverNumber of packets with RX overrun.   
   if_packets        rx: Number of received packets   
   if_tx_octets Number of transmitted bytes   
   if_dropped       tx: Number of packets dropped by TX   
   if_errors           tx: Total number of transmit errors, greater than or equal to the sum of the TX errors above.   
   if_packets        tx: Number of transmitted packets   
 if_packets     rx:1_to_64_packetsThe total number of packets (including bad  packets) received that were 64 octets in length (excluding framing bits but including FCS octets). supported in OvS v2.6+ and dpdk ports only 
 if_packets     rx:65_to_127_packetsThe total number of packets (including bad packets) received that were between 128 and 255 octets in length inclusive (excluding framing bits but including FCS octets). supported in OvS v2.6+ and dpdk ports only 
  if_packets     rx:128_to_255_packetsThe total number of packets (including bad packets) received that were between 256 and 511 octets in length inclusive (excluding framing bits but including FCS octets). supported in OvS v2.6+ and dpdk ports only 
  if_packets     rx:256_to_511_packetsThe total number of packets (including badpackets) received that were between 512 and 1023 octets in length inclusive (excluding framing bits but including FCS octets). supported in OvS v2.6+ and dpdk ports only 
  if_packets     rx:512_to_1023_packetsThe total number of packets (including bad packets) received that were between 1024 and 1518 octets in length inclusive (excluding framing bits but including FCS octets). supported in OvS v2.6+ and dpdk ports only 
  if_packets     rx:1024_to_1522_packetsThe total number of packets (including bad packets) received that were between 1523 and max octets in length inclusive (excluding framing bits but including FCS octets). supported in OvS v2.6+ and dpdk ports only 
  if_packets     rx:1523_to_max_packetsThe total number of packets (including bad packets) received that were between 1523 and max octets in length inclusive (excluding framing bits but including FCS octets). supported in OvS v2.6+ and dpdk ports only 
 if_packets     tx:1_to_64_packetsThe total number of packets transmitted that were 64 octets in length. supported in OvS v2.6+ and dpdk ports only 
  if_packets     tx:65_to_127_packetsThe total number of packets received that were between 65 and 127 octets in length inclusive supported in OvS v2.6+ and dpdk ports only 
  if_packets     tx:128_to_255_packetsThe total number of packets received that were between 128 and 255 octets in length inclusive supported in OvS v2.6+ and dpdk ports only 
  if_packets     tx:256_to_511_packetsThe total number of packets received that were between 256 and 511 octets in length inclusive supported in OvS v2.6+ and dpdk ports only 
  if_packets     tx:512_to_1023_packetsThe total number of packets received that were between 512 and 1023 octets in length inclusive supported in OvS v2.6+ and dpdk ports only 
  if_packets     tx:1024_to_1522_packetsThe total number of packets received that were between 1024 and 1518 octets in length inclusive supported in OvS v2.6+ and dpdk ports only 
  if_packets     tx:1523_to_max_packetsThe total number of packets received that were between 1523 and max octets in length inclusive supported in OvS v2.6+ and dpdk ports only 
  if_multicasttx_multicast_packetsThe number of good packets transmitted that were directed to a multicast.
Note: that this number does not include packets directed to the broadcast address 
 supported in OvS v2.6+ and dpdk ports only 
 if_packets    rx:broadcast_packetsThe total number of packets (including bad packets, broadcast packets, and multicast packets) received. supported in OvS v2.6+ and dpdk ports only 
 if_packets    tx:broadcast_packetsThe number of good packets transmitted that were directed to the broadcast address. supported in OvS v2.6+ and dpdk ports only 
 if_rx_errorsrx_undersized_errorsThe total number of packets received that were less than 64 octets long (excluding framing bits, but including FCS octets) and were otherwise well formed. supported in OvS v2.6+ and dpdk ports only 
 if_rx_errorsrx_oversize_errorsThe total number of packets received that were longer than max octets (excluding framing bits, but including FCS octets) and were otherwise well formed. supported in OvS v2.6+ and dpdk ports only 
 if_rx_errorsrx_fragmented_errors

The total number of packets received that were less than 64 octets in length (excluding framing bits but including FCS octets) and had either a bad Frame Check Sequence (FCS) with an integral number of octets (FCS Error) or a bad FCS with a non-integral number of octets (Alignment Error).

Note: that it is entirely normal for rx_fragmented_errors to increment.  This is because it counts both runts (which are normal occurrences due to collisions) and noise hits
 supported in OvS v2.6+ and dpdk ports only 
 if_rx_errorsrx_jabber_errorsThe total number of jabber packets received that had either a bad Frame Check Sequence (FCS) with an integral number of octets (FCS Error) or a bad FCS with a non-integral number of octets (Alignment Error). supported in OvS v2.6+ and dpdk ports only 

Hugepages

 

(A read plugin that retrieves the number of available and free hugepages on a platform as well as what is available in terms of hugepages per socket.)


 bytesusedNumber of used hugepages in bytes total/pernode/both

Virtual memory makes it easy for several processes to share memory [TM1] .

Each process has its own virtual address space, which is mapped to physical memory by the operating system” [TM2] . The process views the virtual memory address space as a contiguous/linear address space, but in reality – the virtual addresses need to be mapped to physical addresses, this is typically done by the Memory Management Unit (MMU) on the CPU.

“There are two ways to enable the system to manage large amounts of memory:

  • Increase the number of page table entries in the hardware memory management unit
  • Increase the page size/use huge pages à to reduce the number of lookups

The first method is expensive, since the hardware memory management unit in a modern processor only supports hundreds or thousands of page table entries. Additionally, hardware and memory management algorithms that work well with thousands of pages (megabytes of memory) may have difficulty performing well with millions (or even billions) of pages. This results in performance issues: when an application needs to use more memory pages than the memory management unit supports, the system falls back to slower, software-based memory management, which causes the entire system to run more slowly.

Huge pages are blocks of memory that come in 2MB and 1GB sizes. The page tables used by the 2MB pages are suitable for managing multiple gigabytes of memory, whereas the page tables of 1GB pages are best for scaling to terabytes of memory” [TM3] 

More info on virtual memory and TLB lookups can be found:

  1. http://www.tldp.org/LDP/tlk/mm/memory.html
  2. https://lwn.net/Articles/253361/
 bytesfreeNumber of free hugepages in bytes  
 vmpage_numberusedNumber of used hugepages in numbers  
 vmpage_numberfreeNumber of free hugepages in numbers  
 percentusedNumber of used hugepages in percent  
 percentfreeNumber of free hugepages in percent  

processes

(A read plugin that  collects the number of processes, grouped by their state (e. g. running, sleeping, zombies, etc.). In addition to that, it can select detailed statistics about selected processes, grouped by name.)

 fork_rate the number of threads created since the last reboot The information comes mainly from /proc/PID/status, /proc/PID/psinfo and /proc/PID/usage. 
 ps_stateblockedthe number of processes in a blocked state https://collectd.org/wiki/index.php/Plugin:Processes 
 ps_statepagingthe number of processes in a paging state http://man7.org/linux/man-pages/man5/proc.5.html 
 ps_staterunningthe number of processes in a running state   
 ps_statesleepingthe number of processes in a sleeping state   
 ps_statestoppedthe number of processes in a stopped state   
 ps_statezombiesthe number of processes in a Zombie state   
Host only

Libvirt

(A read plugin that uses virtualization API libvirt to gather statistics about virtualized guests on a system directly from the hypervisor, without a need to install collectd instance on the guest.)

 disk_octetsDISKnumber of read/write bytes as unsigned long long.   
 disk_opsDISKnumber of read/write requests    
 disk_timeflush-DISKtotal time spend on cache reads/writes in nano-seconds   
 if_droppedINTERFACEpackets dropped on rx/tx as unsigned long long   
 if_errorsINTERFACErx/tx errors as unsigned long long   
 if_octetsINTERFACEbytes received/transmitted as unsigned long long   
 if_packetsINTERFACEpackets received/transmitted as unsigned long long   
 memoryactual_balloonResident Set Size of the process running the domain. This value is in kB https://libvirt.org/html/libvirt-libvirt-domain.html#virDomainMemoryStatStruct 
 memoryrssHow much the balloon can be inflated without pushing the guest system to swap, corresponds to 'Available' in /proc/meminfo   
 memoryswap_inThe total amount of memory written out to swap space (in kB).   
 memorytotalthe memory in KBytes used by the domain   
 virt_cpu_total the CPU time used in nanoseconds   
 virt_vcpuVCPU_NRthe CPU time used in nanoseconds per cpu   
 cpu_affinityvcpu_NR-cpu_NRpinning of domain VCPUs to host physical CPUs. Value stored is a boolean. 
 job_stats*Information about progress of a background/completed job on a domain. Number of metrics depend on job type. Check API documentation for more information: virDomainGetJobStats 
 disk_errorDISK_NAMEDisk error code Metric isn’t dispatched for disk with no errors 
 percentvirt_cpu_totalCPU utilization in percentage per domain   
 perf*Performance monitoring events Number of metrics depends on libvirt API version. Following perf metric are avilable in libvirt API version 2.4. To collectd perf metric they must be enabled in domain and supported by the platform. 
 perfperf_cmtusage of l3 cache in bytes by applications running on the platform   
 perfperf_ mbmttotal system bandwidth from one level of cache   
 perfperf_ mbmlbandwidth of memory traffic for a memory controller   
 perfperf_cpu_cyclesthe count of cpu cycles (total/elapsed)   
 perfperf_instructionsthe count of instructions by applications running on the platform   
 perfperf_cache_referencesthe count of cache hits by applications running on the platform   
 perfperf_cache_missesthe count of cache misses by applications running on the platform   
 ps_cputime physical user/system cpu time consumed by the hypervisor   
 total_requestsflush-DISKtotal flush requests of the block device   
 total_time_in_msflush-DISKtotal time spend on cache flushing in milliseconds   

RDT

(A read plugin that provides the last level cache utilization and memory bandwidth utilization)


NOTE: it's recommended to configure the interval for this plugin as 1 second

 ipc Number of instructions per clock per core group per core group

A higher IPC means that the processor can get more work done per unit time, which generally translates to faster application performance [0]

"Ideally every instruction a CPU gets should be read, executed and finished in one cycle, however that is never the case. The processor has to take the instruction, decode the instruction, gather the data (depends on where the data is), perform work on the data, then decide what to do with the result. Moving has never been more complicated, and the ability for a processor to hide latency, pre-prepare data by predicting future events or keeping hold of previous events for potential future use is all part of the plan. All the meanwhile there is an external focus on making sure power consumption is low and the frequency of the processor can scale depending on what the target device actually is". [1]

0 https://www.nextplatform.com/2016/03/31/examining-potential-hpc-benefits-new-intel-xeon-processors/

1 http://www.anandtech.com/show/9482/intel-broadwell-pt2-overclocking-ipc/3

 memory_bandwidthlocal

Local Memory Bandwidth in Bytes per second, for a specified processor (socket), Resource Monitoring ID (RMID), and a given measurement time over the preceding measurement interval.

RMID may be mapped to Threads, Application processes, or VMs by the host OS or hypervisor.

Local refers to Memory Bandwidth used by RMID within the specified processor (socket).

 Question is whether this metric is always reported in units of MegaBytes per second.

https://software.intel.com/en-us/articles/memory-bandwidth-monitoring-proof-points

 memory_bandwidthremote 

Remote Memory Bandwidth in Bytes per second, for a specified processor (socket), Resource Monitoring ID (RMID), and a given measurement time over the preceding measurement interval.

 

RMID may be mapped to Threads, Application processes, or VMs by the host OS or hypervisor.

Remote refers to Memory Bandwidth used by RMID external to the specified processor (socket), such as the other processor in a pair.
  
 bytesllcLast Level Cache occupancy in bytes, for a specified processor (socket), Resource Monitoring ID (RMID), and a given measurement time over the preceding measurement interval.  

RMID and Class of Service (CLOS) limits may be mapped to Threads, Application processes, or VMs by the host OS or hypervisor.

 

Question if units are in bytes or kibibits.

Al Morton after looking into this, I found that kilobytes is what is used for the cache.
In collectd we are reporting it simply in bytes. I updated the 4th column to reflect that as well as the definition.

 
Host/guest

dpdkstats

(A read plugin that retrieve stats from the DPDK extended

 

NIC stats API.)

 deriverx_l3_l4_xsum_error Number of receive IPv4, TCP, UDP or SCTP XSUM errors.   
 errors


flow_director_filter_add_errorsNumber of failed added filters compatible with DPDK 16.04, 16.07 (based on ixgbe, vhost support will be enabled in DPDK 16.11) 
 flow_director_filter_remove_errorsNumber of failed removed filters   
 mac_local_errorsNumber of faults in the local MAC.   
 mac_remote_errorsNumber of faults in the remote MAC.   
 if_rx_dropped


rx_fcoe_droppedNumber of Rx packets dropped due to lack of descriptors.   
 rx_mac_short_packet_droppedNumber of MAC short packet discard packets received.   
 rx_management_droppedNumber of management packets dropped. This register counts the total number of packets received that pass the management filters and then are dropped because the management receive FIFO is full. Management packets include any packet directed to the manageability console (such as RMCP and ARP packets).   
 rx_priorityX_droppedNumber of dropped packets received per UP where X is 0 to 7 
 if_rx_errors














rx_crc_errorsCounts the number of receive packets with CRC errors. In order for a packet to be counted in this register, it must be 64 bytes or greater (from <Destination Address> through <CRC>, inclusively) in length.   
 rx_errorsNumber of errors received   
 rx_fcoe_crc_errorsFC CRC Count.    
  Count the number of packets with good Ethernet CRC and bad FC CRC   
 rx_fcoe_mbuf_allocation_errorsNumber of fcoe Rx packets dropped due to lack of descriptors.   
 rx_fcoe_no_direct_data_placement    
 rx_fcoe_no_direct_data_placement_ext_buff    
 rx_fragment_errorsNumber of receive fragment errors (frame shorted than 64 bytes from <Destination Address> through <CRC>, inclusively) that have bad CRC (this is slightly different from the Receive Undersize Count register).    
 rx_illegal_byte_errorsCounts the number of receive packets with illegal bytes errors (such as there is an illegal symbol in the packet).   
 rx_jabber_errorsNumber of receive jabber errors. This register counts the number of received packets that are greater than maximum size and have bad CRC (this is slightly different from the Receive Oversize Count register). The packets length is counted from <Destination Address> through <CRC>, inclusively.    
 rx_length_errorsNumber of packets with receive length errors. A length error occurs if an incoming packet length field in the MAC header doesn't match the packet length.   
 rx_mbuf_allocation_errorsNumber of Rx packets dropped due to lack of descriptors.   
 rx_oversize_errorseceive Oversize Error. This register counts the number of received frames that are longer than maximum size as defined by MAXFRS.MFS (from <Destination Address> through <CRC>, inclusively) and have valid CRC.    
 rx_priorityX_mbuf_allocation_errorsNumber of received packets per UP dropped due to lack of descriptors. where X is 0 to 7 
 rx_q0_errorsNumber of errors received for the queue. if more queues are allocated then you get the errors per Queue 
 rx_undersize_errorsReceive Undersize Error. This register counts the number of received frames that are shorter than minimum size (64 bytes from <Destination Address> through <CRC>, inclusively), and had a valid CRC.   
 if_rx_octets



rx_error_bytesCounts the number of receive packets with error bytes (such as there is an error symbol in the packet). This registers counts all packets received, regardless of L2 filtering and receive enablement. bug - will move this to errors 
 rx_fcoe_bytesnumber of received fcoe bytes   
 rx_good_bytesGood octets/bytes received count. This register includes bytes received in a packet from the <Destination Address> field through the <CRC> field, inclusively.   
 rx_q0_bytesNumber of bytes received for the queue. per queue 
 rx_total_bytesTotal received octets. This register includes bytes received in a packet from the <Destination Address> field through the <CRC> field, inclusively.   
 if_rx_packets


















rx_broadcast_packetsNumber of good (non-erred) broadcast packets received.   
 rx_fcoe_packetsNumber of FCoE packets posted to the host. In normal operation (no save bad frames) it equals to the number of good packets.   
 rx_flow_control_xoff_packetsNumber of XOFF packets received. This register counts any XOFF packet whether it is a legacy XOFF or a priority XOFF. Each XOFF packet is counted once even if it is designated to a few priorities.   
 rx_flow_control_xon_packetsNumber of XON packets received. This register counts any XON packet whether it is a legacy XON or a priority XON. Each XON packet is counted once even if it is designated to a few priorities.   
 rx_good_packetsNumber of good (non-erred) Rx packets (from the network).   
 rx_management_packetsNumber of management packets received. This register counts the total number of packets received that pass the management filters. Management packets include RMCP and ARP packets. Any packets with errors are not counted, except for the packets that are dropped because the management receive FIFO is full are counted.   
 rx_multicast_packetsNumber of good (non-erred) multicast packets received (excluding broadcast packets). This register does not count received flow control packets.    
 rx_priorityX_xoff_packetsNumber of XOFF packets received per UP where X is 0 to 7 
 rx_priorityX_xon_packetsNumber of XON packets received per UP where X is 0 to 7 
 rx_q0_packetsNumber of packets received for the queue. per queue 
 rx_size_1024_to_max_packetsNumber of packets received that are 1024-max bytes in length (from <Destination Address> through <CRC>, inclusively). This registers does not include received flow control packets. The maximum is dependent on the current receiver configuration and the type of packet being received. If a packet is counted in receive oversized count, it is not counted in this register. Due to changes in the standard for maximum frame size for VLAN tagged frames in 802.3, packets can have a maximum length of 1522 bytes.   
 rx_size_128_to_255_packetsNumber of packets received that are 128-255 bytes in length (from <Destination Address> through <CRC>, inclusively).   
 rx_size_256_to_511_packetsNumber of packets received that are 256-511 bytes in length (from <Destination Address> through <CRC>, inclusively).   
 rx_size_512_to_1023_packetsNumber of packets received that are 512-1023 bytes in length (from <Destination Address> through <CRC>, inclusively).   
 rx_size_64_packetsNumber of good packets received that are 64 bytes in length (from <Destination Address> through <CRC>, inclusively).   
 rx_size_65_to_127_packetsNumber of packets received that are 65-127 bytes in length (from <Destination Address> through <CRC>, inclusively)   
 rx_total_missed_packetsthe total number of rx missed packets, that is is a packet that was correctly received by the NIC but because it was out of descriptors and internal memory, the packet had to be dropped by the NIC itself   
 rx_total_packetsNumber of all packets received. This register counts the total number of all packets received. All packets received are counted in this register, regardless of their length, whether they are erred, but excluding flow control packets.   
 rx_xoff_packetsNumber of XOFF packets received. Sticks to 0xFFFF. XOFF packets can use the global address or the station address. This register counts any XOFF packet whether it is a legacy XOFF or a priority XOFF. Each XOFF packet is counted once even if it is designated to a few priorities. If a priority FC packet contains both XOFF and XON, only this counter is incremented.   
 rx_xon_packetsNumber of XON packets received. XON packets can use the global address, or the station address. This register counts any XON packet whether it is a legacy XON or a priority XON. Each XON packet is counted once even if it is designated to a few priorities. If a priority FC packet contains both XOFF and XON, only the LXOFFRXCNT counter is incremented.   
 if_tx_errorstx_errorsTotal number of TX error packets   
 if_tx_octets

tx_fcoe_bytesNumber of fcoe bytes transmitted   
 tx_good_bytescounter of successfully transmitted octets. This register includes transmitted bytes in a packet from the <Destination Address> field through the <CRC> field, inclusively.   
 tx_q0_bytesNumber of bytes transmitted by the queue. per queue 
 if_tx_packets

















tx_broadcast_packetsNumber of broadcast packets transmitted count. This register counts all packets, including standard packets, secure packets, FC packets and manageability packets   
 tx_fcoe_packetsNumber of fcoe packets transmitted   
 tx_flow_control_xoff_packetsLink XOFF Transmitted Count   
 tx_flow_control_xon_packetsLink XON Transmitted Count   
 tx_good_packetsNumber of good packets transmitted   
 tx_management_packetsNumber of management packets transmitted.   
 tx_multicast_packetsNumber of multicast packets transmitted. This register counts the number of multicast packets transmitted. This register counts all packets, including standard packets, secure packets, FC packets and manageability packets.   
 tx_priorityX_xoff_packetsNumber of XOFF packets transmitted per UP where X is 0 to 7 
 tx_priorityX_xon_packetsNumber of XON packets transmitted per UP where X is 0 to 7 
 tx_q0_packetsNumber of packets transmitted for the queue. A packet is considered as transmitted if it is was forwarded to the MAC unit for transmission to the network and/or is accepted by the internal Tx to Rx switch enablement logic. Packets dropped due to anti-spoofing filtering or VLAN tag validation (as described in Section 7.10.3.9.2) are not counted. per queue 
 tx_size_1024_to_max_packetsNumber of packets transmitted that are 1024 or more bytes in length (from <Destination Address> through <CRC>, inclusively). This register counts all packets, including standard packets, secure packets, and manageability packets.   
 tx_size_128_to_255_packetsNumber of packets transmitted that are 128-255 bytes in length (from <Destination Address> through <CRC>, inclusively). This register counts all packets, including standard packets, secure packets, and manageability packets   
 tx_size_256_to_511_packetsNumber of packets transmitted that are 256-511 bytes in length (from <Destination Address> through <CRC>, inclusively). This register counts all packets, including standard packets, secure packets, and manageability packets.   
 tx_size_512_to_1023_packetsNumber of packets transmitted that are 512-1023 bytes in length (from <Destination Address> through <CRC>, inclusively). This register counts all packets, including standard packets, secure packets, and manageability packets.   
 tx_size_64_packetsNumber of packets transmitted that are 64 bytes in length (from <Destination Address> through <CRC>, inclusively). This register counts all packets, including standard packets, secure packets, FC packets, and manageability packets.   
 tx_size_65_to_127_packetsNumber of packets transmitted that are 65-127 bytes in length (from <Destination Address> through <CRC>, inclusively). This register counts all packets, including standard packets, secure packets, and manageability packets.   
 tx_total_packetsNumber of all packets transmitted. This register counts the total number of all packets transmitted. This register counts all packets, including standard packets, secure packets, FC packets, and manageability packets.   
 tx_xoff_packetsNumber of XOFF packets transmitted   
 tx_xon_packetsNumber of XON packets transmitted   
 operations


flow_director_added_filtersThis field counts the number of added filters to the flow director filters logic.   
 flow_director_matched_filtersThis field counts the number of matched filters to the flow director filters logic.   
 flow_director_missed_filtersThis field counts the number of missed filters to the flow director filters logic.   
 flow_director_removed_filtersThis field counts the number of removed filters from the flow director filters logic.   

mcelog (RAS memory)

 

A read plugin that uses mcelog to check for memory Machine Check Exceptions and sends the stats for reported exceptions


 errors




corrected_memory_errorsThe total number of hardware errors that were corrected by the hardware (e.g. using a single bit data corruption that was correctible using ECC). These errors do not require immediate software actions, but are still reported for accounting and predictive failure analysis. Memory (RAM) errors are among the most common errors in typical server systems. They also scale with the amount of memory: the more memory the more errors. In addition large clusters of computers with tens or hundreds (or sometimes thousands) of active machines increase the total error rate of the system. 
 uncorrected_memory_errorthe total number of uncorrected hardware errors detected by the hardware. Data corruption has occurred. These errors require software reaction. http://www.mcelog.org/memory.html 
 corrected_memory_errors_in_%sThe total number of hardware errors that were corrected by the hardware in a certain period of time where %s is a timed period like 24 hours 
    http://www.mcelog.org/memory.html  
 uncorrected_memory_errors_in_%sthe total number of uncorrected hardware errors detected by the hardware in a certain period of time where %s is a timed period like 24 hours 
    http://www.mcelog.org/memory.html  
Host

IPMI

A read plugin that reports platform thermals, voltages, fan speed, current, flow, power etc.


(specific per BMC) so these will change depending on what's supported by the BMC.
This is en example for S2600WT2R platform

 percent MTT CPU2 (Memory Throttling sensor) IPMI defines many types of sensors, but groups them into two
main categories: Threshold and discrete.

Threshold sensors are “analog”, they have continuous (or mostly continuous) readings. Things like fans speed, voltage, or temperature.

Discrete sensors have a set of binary readings that may each be independently zero or one. In some sensors, these may be independent. For instance, a power supply may have both an external power failure and a predictive failure at the same time. In other cases they may be mutually exclusive. For instance, each
bit may represent the initialization state of a piece of software.
 The IPMI plugin supports analog sensors of type voltage,  temperature,  fan and  current + analog sensors that have VALUE type WATTS, CFM and percentage (%).

http://openipmi.sourceforge.net/IPMI.pdf 

https://www.intel.com/content/dam/support/us/en/documents/motherboards/server/s5400sf/sb/s5400sf_tps_r2_02.pdf
 
  MTT CPU1 (Memory Throttling sensor)  
  P2 Therm Ctrl %  
  P1 Therm Ctrl %  
  PS1 Curr Out %  
 voltage BB +3.3V Vbat  
  BB +12.0V  
 temperature Agg Therm Mgn 1  
  DIMM Thrm Mrgn 4  
  DIMM Thrm Mrgn 3  
  DIMM Thrm Mrgn 2  
  DIMM Thrm Mrgn 1  
  P2 DTS Therm Mgn  
  P1 DTS Therm Mgn  
  P2 Therm Ctrl %  
  P1 Therm Ctrl %  
  P2 Therm Margin  
  P1 Therm Margin  
  PS1 Temperature  
  LAN NIC Temp  
  Exit Air Temp  
  HSBP 1 Temp  
  I/O Mod Temp  
  BB Lft Rear Temp  
  BB Rt Rear Temp  
  BB BMC Temp  
  SSB Temp  
  Front Panel Temp  
  BB P2 VR Temp  
  BB P1 VR Temp  
 fan System Fan 6B  
  System Fan 6A  
  System Fan 5B  
  System Fan 5A  
  System Fan 4B  
  System Fan 4A  
  System Fan 3B  
  System Fan 3A  
  System Fan 2B  
  System Fan 2A  
  System Fan 1B  
  System Fan 1A  
  CFM System Airflow  
  watts PS1 Input Power  
Host

intel_pmu

 

A read plugin that collects performance monitoring events supported by Intel Performance Monitoring Units (PMUs). The PMU is hardware built inside a processor to measure its performance parameters such as instruction cycles, cache hits, cache misses, branch misses and many others. Performance monitoring events provide facilities to characterize the interaction between programmed sequences of instructions and microarchitectural sub-systems.


 countercpu-cycles[Hardware event] The types of events are:

Hardware Events: These instrument low-level processor activity based on CPU performance counters. For example, CPU cycles, instructions retired, memory stall cycles, level 2 cache misses, etc. Some will be listed as Hardware Cache Events.
Software Events: These are low level events based on kernel counters. For example, CPU migrations, minor faults, major faults, etc.

http://www.brendangregg.com/perf.html#Events 
 
 instructions  
 cache-references  
 cache-misses  
 branch-instructionsORbranches  
 branch-misses  
 bus-cycles  
 cpu-clock[Software event]  
 task-clock  
 page-faultsORfaults  
 minor-faults  
 major-faults  
 context-switchesORcs  
 cpu-migrationsORmigrations  
 alignment-faults  
 emulation-faults  
 L1-dcache-loads[Hardware cache event]  
 L1-dcache-load-misses  
 L1-dcache-stores  
 L1-dcache-store-misses  
 L1-dcache-prefetches  
 L1-dcache-prefetch-misses  
 L1-icache-loads  
 L1-icache-load-misses  
 L1-icache-prefetches  
 L1-icache-prefetch-misses  
 LLC-loads  
 LLC-load-misses  
 LLC-stores  
 LLC-store-misses  
 LLC-prefetches  
 LLC-prefetch-misses  
 dTLB-loads  
 dTLB-load-misses  
 dTLB-stores  
 dTLB-store-misses  
 dTLB-prefetches  
 dTLB-prefetch-misses  
 iTLB-loads  
 iTLB-load-misses  
 branch-loads  
 branch-load-misses  
HostOVS PMD stats

main thread

counter

emc hits

Number of packets hitting Exact Match Cache.

  

Plugin instance: pmd thread can be different combinations of <numa_id> and <core_id>.

e.g: ovs_pmd_stats-pmd_thread_numa_id_0_core_id_5

pmd thread _ numa_id #value _ core_id #value

megaflow hits

Number of packets hitting Megaflow Cache.  
avg. subtable lookups per hit

Average number of subtable lookups for every hit.

  
miss

Number of packets not matching any existing flow.

  
lost

Number of packets destined for user space process but subsequently dropped before reaching userspace.

  

polling cycles

Number of cycles used for polling packets.

  

processing cycles

Number of cycles used for processing incoming packets.

  
avg cycles per packet

Average number of cycles per packet.

  

avg processing cycles per packet

Average number of processing cycles per packet.  
         

Events

NOTE: Collectd can generate events based on thresholds for any of the metrics reported in the table above. For more info please see: https://collectd.org/documentation/manpages/collectd.conf.5.shtml#threshold_configuration 

Where collectd is runningPluginTypeType InstanceSeverityDescriptioncomment
host/guestovs_eventsgaugelink_statusWarning on Link Status DownLink status of the OvS interface: UP or DOWN
Severity will be configurable by the end user
 
OKAY on link Status Up
host/guest

dpdk_events

 

A read plugin that retrieves events (like link status changes) from OVS.


 link_statusWarning on Link Status Down, OKAY on link status upLink status of the OvS interface: UP or DOWN
Severity will be configurable by the end user
Depending on plugin configuration, can be dispatched as a metric or event.
 keep_aliveOKAY: if core status is ALIVE, UNUSED, DOZING, SLEEP
Warning: if core status is MISSING
Failure: if core status is DEAD or GONE
Reflects the state of DPDK packet processing coresprotects against packet processing core failures for DPDK --> no slient packet drops. Depending on plugin configuration, can be dispatched as a metric or event.
host

pcie

 

A read plugin that monitors PCIe standard and advanced errors and sends notifications about those errors


pcie_errorcorrectableNotification (Warning) in case of PCIe correctable error occurrence. Message contains short error description.

Correctable Errors include:
Receiver Error Status
Bad TLP Status
Bad DLLP Status
REPLAY_NUM Rollover
Replay Timer Timeout
Advisory Non-Fatal
Corrected Internal
Header Log Overflow

 

Uncorrectable Errors include:
Data Link Protocol
Surprise Down
Poisoned TLP
Flow Control Protocol
Completion Timeout
Completer Abort
Unexpected Completion
Receiver Overflow
Malformed TLP
ECRC Error Status
Unsupported Request
ACS Violation
Internal
MC blocked TLP
Atomic egress blocked
TLP prefix blocked

 
fatalNotification (Failure) in case of PCIe uncorrectable fatal error occurrence. Message contains short error description.
non_fatalNotification (Warning) in case of PCIe uncorrectable non-fatal error occurrence. Message contains short error description.
host

mcelog (RAS memory)

 

A read plugin that uses mcelog to check for memory Machine Check Exceptions and sends notifications for reported exceptions


errors 

Warning for Corrected Memory Errors

Failure for Uncorrected Memory Errors

Failure on failure to connect to the mcelog socket/ if connection is lost

OK on connection to mcelog socket

Reports Corrected and Uncorrected DIMM Failures
host

IPMI

 

A read plugin that (IPMI) System Event Log (SEL) and sends appropriate notifications based on monitored SEL events and threshold events.


  OKAY - upper non-criticalEach IPMI sensor may have six different thresholds: upper non-recoverable upper critical upper non-critical lower non-critical lower critical lower non-recoverableYou may have events on a threshold sensor by specifying values (called thresholds) where you want the sensor to report an event. Then you can enable the events for the specific thresholds. Not all sensors support all thresholds, some cannot have their events enabled and others cannot have them disabled. The capabilities of a sensor may all be queried by the user to determine what it can do. When the value of the sensor goes outside the threshold an event may be generated. An event may be generated when the value goes back into the threshold
OKAY - lower non-critical
WARNING- lower critical
WARNING - upper critical
FAILURE - upper non-recoverable
FAILURE - lower non-recoverable
discrete sensor status changes are also reported out via OKAY, WARNING and FAILURE notifications.

Examples of discrete sensors can be found under the "IPMI Sensors for S2600WT2R" tab
host

mcelog RAS System, CPU, QPI, OI

(specific to a Platform) so these will change depending on what's supported by the Platform.


 

A read plugin that uses mcelog to check for cpu, IO, QPI or system Machine Check Exceptions and sends the stats for reported exceptions


  WARNING - Correctable errors
FAILURE - Uncorrectable Errors
Servers based on Intel® Architecture, are generally designed for use in mission critical environments. Reliability, Availability and Serviceability (RAS) features, are integrated into the servers to address the error handling and memory mirroring and sparing required by these environments.

The goal of this  feature is to expose the RAS features provided by the Broadwell or newer platfrom to higher level fault management applications.

The Features to be exposed fall under the following catagories:
Reliability Features:
-System attributes to ensure Data integrity.
-capability to prevent, detect, correct and contain faults over a given time interval.
Availability Features:
-System attributes to help stay operational in the presence of faults in the system.
-Capability to map out failed units, ability to operate in a degraded mode.
Serviceability Features:
-System attributes to help system service, repair.
-Capability to identify failed units, and facilitates repair.


Generic Error Handling
The Silicon supports corrected, uncorrected (recoverable, unrecoverable), fatal and catastrophic error types.
Corrected Errors
Errors that are corrected by either hardware or software, corrected error information is used in predictive failure analysis by the OS.
MCA Banks corrected errors except selected memory corrected errors are handled directly by the OS. HASWELL-EP PROCESSOR triggers CMCI for the corrected errors, on CMCI OS can read the MCA Banks and collect error status. All the other platform related corrected errors can either be ignored or can be logged into BMC SEL based on platform policy.
Memory Corrected Errors
Memory corrected errors such as mirror fail over, memory read errors can be configured to trigger SMI using BIOS setup options. On memory mirror fail over BIOS logs the error for the OS as per the UEFI error record format. On memory read errors, BIOS does the following memory RAS operations in the order to correct the error.
Rank Sparing
SDDC/Device tagging
UnCorrected Non Fatal Errors
Errors that are not corrected by hardware, in general these errors trigger machine check exception and in turn triggers SMI. BIOS SMI handler logs these error information, clear the error status and pass the error log to OS. OS can recover from the error, in cases where the recovery is not an option, can trigger a system reset.
Uncorrected Fatal Errors
Errors that are neither corrected by hardware nor recovered by the s/w, the system is not in a reliable state and needs a reset to bring it back up to normal operation. In most fatal error conditions, BIOS cannot log errors before the system reset happens. All the Error status registers are sticky on the reset, BIOS collects all these information in the next boot, creates error record and pass it on the OS.
Error Logging

Example Errors are provided in the comments tab are for the Purley platform
/* See IA32 SDM Vol3B Chapter 16*/
Integrated Memory Controller Machine Check Errors
"Address parity error",
"HA write data parity error",
"HA write byte enable parity error",
"Corrected patrol scrub error",
"Uncorrected patrol scrub error",
"Corrected spare error",
"Uncorrected spare error",
"Any HA read error",
"WDB read parity error",
"DDR4 command address parity error",
"Uncorrected address parity error"
"Unrecognized request type",
"Read response to an invalid scoreboard entry",
"Unexpected read response",
"DDR4 completion to an invalid scoreboard entry",
"Completion to an invalid scoreboard entry",
"Completion FIFO overflow",
"Correctable parity error",
"Uncorrectable error",
"Interrupt received while outstanding interrupt was not ACKed",
"ERID FIFO overflow",
"Error on Write credits",
"Error on Read credits",
"Scheduler error",
"Error event",
"MscodDataRdErr",
"Reserved",
"MscodPtlWrErr",
"MscodFullWrErr",
"MscodBgfErr",
"MscodTimeout",
"MscodParErr",
"MscodBucket1Err"

Interconnect(QPI) Machine Check Errors
"UC Phy Initialization Failure",
"UC Phy detected drift buffer alarm",
"UC Phy detected latency buffer rollover",
"UC LL Rx detected CRC error: unsuccessful LLR: entered abort state",
"UC LL Rx unsupported or undefined packet",
"UC LL or Phy control error",
"UC LL Rx parameter exchange exception",
"UC LL detected control error from the link-mesh interface",
"COR Phy initialization abort",
"COR Phy reset",
"COR Phy lane failure, recovery in x8 width",
"COR Phy L0c error corrected without Phy reset",
"COR Phy L0c error triggering Phy Reset",
"COR Phy L0p exit error corrected with Phy reset",
"COR LL Rx detected CRC error - successful LLR without Phy Reinit",
"COR LL Rx detected CRC error - successful LLR with Phy Reinit"
"Phy Control Error",
"Unexpected Retry.Ack flit",
"Unexpected Retry.Req flit",
"RF parity error",
"Routeback Table error",
"unexpected Tx Protocol flit (EOP, Header or Data)",
"Rx Header-or-Credit BGF credit overflow/underflow",
"Link Layer Reset still in progress when Phy enters L0",
"Link Layer reset initiated while protocol traffic not idle",
"Link Layer Tx Parity Error"
Internal Machine Check Errors
"No Error",
"MCA_DMI_TRAINING_TIMEOUT",
"MCA_DMI_CPU_RESET_ACK_TIMEOUT",
"MCA_MORE_THAN_ONE_LT_AGENT",
"MCA_BIOS_RST_CPL_INVALID_SEQ",
"MCA_BIOS_INVALID_PKG_STATE_CONFIG",
"MCA_MESSAGE_CHANNEL_TIMEOUT",
"MCA_MSGCH_PMREQ_CMP_TIMEOUT",
"MCA_PKGC_DIRECT_WAKE_RING_TIMEOUT",
"MCA_PKGC_INVALID_RSP_PCH",
"MCA_PKGC_WATCHDOG_HANG_CBZ_DOWN",
"MCA_PKGC_WATCHDOG_HANG_CBZ_UP",
"MCA_PKGC_WATCHDOG_HANG_C3_UP_SF",
"MCA_SVID_VCCIN_VR_ICC_MAX_FAILURE",
"MCA_SVID_COMMAND_TIMEOUT",
"MCA_SVID_VCCIN_VR_VOUT_FAILURE",
"MCA_SVID_CPU_VR_CAPABILITY_ERROR",
"MCA_SVID_CRITICAL_VR_FAILED",
"MCA_SVID_SA_ITD_ERROR",
"MCA_SVID_READ_REG_FAILED",
"MCA_SVID_WRITE_REG_FAILED",
"MCA_SVID_PKGC_INIT_FAILED",
"MCA_SVID_PKGC_CONFIG_FAILED",
"MCA_SVID_PKGC_REQUEST_FAILED",
"MCA_SVID_IMON_REQUEST_FAILED",
"MCA_SVID_ALERT_REQUEST_FAILED",
"MCA_SVID_MCP_VR_ABSENT_OR_RAMP_ERROR",
"MCA_SVID_UNEXPECTED_MCP_VR_DETECTED",
"MCA_FIVR_CATAS_OVERVOL_FAULT",
"MCA_FIVR_CATAS_OVERCUR_FAULT",
"MCA_WATCHDOG_TIMEOUT_PKGC_SLAVE",
"MCA_WATCHDOG_TIMEOUT_PKGC_MASTER",
"MCA_WATCHDOG_TIMEOUT_PKGS_MASTER",
"MCA_PKGS_CPD_UNCPD_TIMEOUT",
"MCA_PKGS_INVALID_REQ_PCH",
"MCA_PKGS_INVALID_REQ_INTERNAL",
"MCA_PKGS_INVALID_RSP_INTERNAL",
"MCA_PKGS_SMBUS_VPP_PAUSE_TIMEOUT",
"MCA_RECOVERABLE_DIE_THERMAL_TOO_HOT"
host

virt

 

A read plugin that uses virtualization API libvirt to gather events about virtualized guests on a system directly from the hypervisor, without a need to install collectd instance on the guest.


domain_state 

OKAY:

  • VIR_DOMAIN_NOSTATE
  • VIR_DOMAIN_RUNNING
  • VIR_DOMAIN_SHUTDOWN
  • VIR_DOMAIN_SHUTOFF
Domain state and reason in a human-readable format. 

WARNING:

  • VIR_DOMAIN_BLOCKED
  • VIR_DOMAIN_PAUSED
  • VIR_DOMAIN_PMSUSPENDED

FAILURE:

  • VIR_DOMAIN_CRASHED
hostvirtfile_system OKAYFile system information (mountpoint, device name, filesystem type, number of aliases, disk aliases)Information stored in metadata. Requires Guest Agent to be installed and configured in VM.

SNMP interface

SNMP interface in collectd provides access to collected metrics using SNMP Agent plugin. This plugin is an AgentX subagent that receives and handles queries from SNMP master agent and returns the metrics collected by "read" (collector) plugins. The plugin handles requests only for OIDs specified in configuration file. To handle SNMP queries the plugin gets data from collectd and translates requested values from collectd's internal format to SNMP format. This plugin is a generic plugin and cannot work without configuration. For more details on configuration file see <https://github.com/collectd/collectd/pull/2105/files#diff-9fc6980794a396e7288e1bd17c59a358>


 
  • No labels

7 Comments

  1. I tried to revise column widths of the "Supported Metrics and Events" Table with no success.  Atlassian docs say it's possible (with Confluence 5.9, I think) https://confluence.atlassian.com/confcloud/blog/2016/05/resizable-table-columns-in-confluence

     

  2. Has anyone looked at how to get this data into a format suitable for collectd?  We need a lot of meta/string data to travel along with the telemetry.

    Collectd doesn't seem to handle strings very well and expects numbers.

     

  3. Aaron Smith can you please clarify "Collectd doesn't seem to handle strings very well and expects numbers" do you mean in the metadata? or something else?

  4. A lot of data.... (smile)

    Would it be helpful to suggest as part of this exercise a sample rate for individual plugins?

      1. Added a note at the top of the page as regards intervals, and the only plugin affected then with an exceptional configuration is RDT... added that interval specifically to the table.

  5. Al Morton thanks for doing the definitions, these look great to me. I added a small comment on llc in the table above