[ZBX-14974] Duplicated performance counter paths Created: 2018 Oct 09 Updated: 2024 Apr 10 Resolved: 2018 Oct 31 |
|
Status: | Closed |
Project: | ZABBIX BUGS AND ISSUES |
Component/s: | Proxy (P), Server (S) |
Affects Version/s: | 3.0.22, 3.4.14, 4.0.0, 4.2.0alpha1 |
Fix Version/s: | 4.0.2rc1, 4.2.0alpha1, 4.2 (plan) |
Type: | Problem report | Priority: | Minor |
Reporter: | Andris Zeila | Assignee: | Michael Veksler |
Resolution: | Fixed | Votes: | 0 |
Labels: | vmware | ||
Remaining Estimate: | Not Specified | ||
Time Spent: | Not Specified | ||
Original Estimate: | Not Specified |
Team: | Team A |
Sprint: | Sprint 45, Sprint 46, Nov 2018 |
Story Points: | 2 |
Description |
Zabbix maps VMware performance counter identifiers to human readable paths in format <group>/<name>[<rollup>]. It might have been enough back then, but now we have two different performance counters that gives the same paths: <PerfCounterInfo xsi:type="PerfCounterInfo"> <key>422</key> <nameInfo> <label>Latency</label> <summary>Percentage of time the virtual machine is waiting to access swapped or compressed memory</summary> <key>latency</key> </nameInfo> <groupInfo> <label>Memory</label> <summary>Memory</summary> <key>mem</key> </groupInfo> <unitInfo> <label>%</label> <summary>Percentage</summary> <key>percent</key> </unitInfo> <rollupType>average</rollupType> <statsType>absolute</statsType> <level>2</level> <perDeviceLevel>3</perDeviceLevel> </PerfCounterInfo> and <PerfCounterInfo xsi:type="PerfCounterInfo"> <key>556</key> <nameInfo> <label>Latency</label> <summary>Percentage of time the virtual machine is waiting to access swapped or compressed memory</summary> <key>latency</key> </nameInfo> <groupInfo> <label>Memory</label> <summary>Memory</summary> <key>mem</key> </groupInfo> <unitInfo> <label>%</label> <summary>Percentage</summary> <key>percent</key> </unitInfo> <rollupType>average</rollupType> <statsType>rate</statsType> <level>2</level> <perDeviceLevel>3</perDeviceLevel> </PerfCounterInfo> The current paths uses groupInfo.Key/nameInfo.Key[rollupType]. Somehow we must add statsType without breaking old keys. One way might be adding it only for duplicated counters, so the first would have the current path - mem/latency[average], the second mem/latency[average,rate]. Another improvement - would be good to initialize/resize counters hashset with correct size to avoid slots reallocation and reduce memory fragmentation. |
Comments |
Comment by Michael Veksler [ 2018 Oct 31 ] |
Available in:
|