The availability data collected by the Oracle Lightweight Availability Collection Tool is stored in the form of datagram within the file system of the monitored host. The availability data is embedded in between XML tags. The Availability datagram can be broadly categorized into two sections:
Monitored System Information
Availability Data
The following is a sample of an Oracle Lightweight Availability Collection Tool Availability datagram file.
<?xml version="1.0" encoding="UTF-8" standalone="no"?> <single_system_availability_results> <systemInfo> <hostName>bs6-s0</hostName> <hostId>83254cb1</hostId> <zoneName>global</zoneName> <timeZone>US/Mountain</timeZone> <sysSerialNumber>unknown</sysSerialNumber> <OSName>SunOS</OSName> <OSVersion>5.10</OSVersion> <cpuArchitecture>sparc</cpuArchitecture> <productType>Serverblade1</productType> <lwactVersion>3.1</lwactVersion> </systemInfo> <event type="epoch" utc="1207784519" timeStamp="Wed Apr 9 17:41:59 2008" up="0" dwnPlnd="0" dwnUnplnd="0" dwnUndef="0" cksum="13c8" /> <event type="boot" utc="1207784519" timeStamp="Wed Apr 9 17:41:59 2008" up="76820" dwnPlnd="0" dwnUnplnd="0" dwnUndef="0" cksum="13e4" /> <event type="panic" utc="1207861339" timeStamp="Thu Apr 10 15:02:19 2008 -06:00" up="0" dwnPlnd="0" dwnUnplnd="1" dwnUndef="0" L1causeCode="Unplanned" L2causeCode="Undefined" L3causeCode="Undefined" wasPlanned="2" cksum="2708" /> <event type="boot" utc="1207861340" timeStamp="Thu Apr 10 15:02:20 2008 -06:00" up="8931" dwnPlnd="0" dwnUnplnd="0" dwnUndef="0" cksum="143b" /> <event type="time" utc="1207870271" timeStamp="Thu Apr 10 17:31:11 2008 -06:00" up="85751" dwnPlnd="0" dwnUnplnd="1" dwnUndef="0" elapsed="85752" totAvail="99.999" adjAvail="99.999" cksum="1c95" /> </single_system_availability_results> |
In this sample datagram file, the epoch and boot events do not have a time zone offset. This might happen if you upgrade the Oracle Lightweight Availability Collection Tool from a pre-3.0 version to a later one. The latest Oracle Lightweight Availability Collection Tool will always contain time zone offset information in the event timestamps.
The information collected in between the tags <systemInfo> and </systemInfo> constitutes the system information section. This section provides details about the monitored host, such as the following:
hostname
hostid
zonename (if present)
timezone of the host
system serial number (if known)
OS name and version
CPU architecture (sparc/X86)
product type
the version of the Oracle Lightweight Availability Collection Tool installed on the host
The Availability Data section contains the availability events (boot, epoch, halt, panic) and its corresponding timestamp. All the availability calculations are done based on the data collected under this section.
This section identifies and describes the important fields of the Availability datagram (listed here in alphabetical order).
Definition
Represented in percentage as: ((Total uptime + Total Planned downtime) /Total elapsed time) * 100
Planned downtime is considered as uptime in this instance; hence, the term adjusted availability.
The duration during which the host was out of run level 3 is considered as downtime (that is, the difference in coordinated universal time (UTC) between the outage event and its corresponding boot event). Downtime is recorded as a part of the outage event (panic/halt). It is decided by the wasPlanned field. wasPlanned can be one of the following designations:
Undefined (value of 0)
Planned (value of 1)
Unplanned (value of 2)
In the sample datagram (above), event #2 is a panic event, and event #3 is its corresponding boot event; the difference in UTC of event #3 and event #2 is the downtime. Therefore, downtime = 1207861340 - 1207861339 (= 1 sec)
Since the wasPlanned flag is 2, the downtime is marked against the field dwnUnplnd (Unplanned downtime)
Represented in percentage as: (Total uptime/Total elapsed time) * 100
The following types of events are recorded in the Availability datagram by the Oracle Lightweight Availability Collection Tool:
epoch
Marks the beginning of event tracking. It is recorded only once in the Availability datagram (at the inception). The UTC of this event marks the inception time of the Oracle Lightweight Availability Collection Tool on the monitored host.
boot
Whenever the host returns to run level 3, a boot event is recorded in the datagram along with the corresponding timestamp.
halt
Whenever the host leaves run level 3 to any other level, a halt event is created with the time of halt being the time the host left run level 3.
panic
If the host encounters an un-natural downing such as system crash, upon the subsequent boot of the host (that is, a return to run level 3), a panic event is recorded where the time of the panic event is the time at which the Oracle Lightweight Availability Collection Tool stopped running.
time
Indicates the last recorded UTC for offline reporting. This event contains the consolidated uptime and downtime information. It also reports the elapsed time (measured as the duration in UTC that the Oracle Lightweight Availability Collection Tool is monitoring this host since inception). Apart from this information, the time event also reports system availability in two forms: Total availability and Adjusted availability.
The difference in UTC between the current outage event and the last event before it, which would be a boot event, is measured as uptime.
In the sample datagram (above), if the uptime field in event #1 (boot event) is calculated as the difference in UTC between event #3 and event #2uptime = 1207861339 - 1207784519 (= 76820 secs)