Compteurs de performance (monitoring)
De Wiki1000
Version du 8 août 2017 à 16:42 par Syfre (discuter | contributions)
Les compteurs de performance sont des compteurs d'évènements internes permettant de suivre l'activité du service.
Il existe trois types de compteur :
- Des compteurs d’évènements qui cumulent le nombre d'évènement (n).
- Des compteurs qui cumulent un nombre d'évènement par seconde (n/s).
- Des compteurs qui indique un nombre d'objets (N)
- Des pourcentages d'utilisation (%)
- Des indicateurs (V)
Nom | Type | usage |
---|---|---|
system.cpu.idle | % | System cpu idle (%) |
process.cpu.usage | % | Process cpu usage (%) |
system.memory.virtualSetSize | V | Memory virtual set size (Mo) |
system.memory.workingSetSize | V | Memory working set size (Mo) |
system.memory.pageFault | n/s | Memory page fault (n/s) |
sql.exec.hit | n/s | Sql execution hit (n/s) |
sql.exec.count | n | Sql execution count (n) |
sql.queries.count | N | Sql query count (n) |
sql.sessions.count | N | Sql session count (n) |
services.hit | n/s | Services hit (n/s) |
services.hit.count | n | Services hit count (n) |
objects.count | N | Framework objects count (n) |
code.exec.hit | n/s | Framework codes hit (n/s) |
code.exec.count | n | Framework codes count (n) |
transactions.hit | n/s | Framework transaction commit hit (n/s) |
transactions.count | n | Framework transaction commit count (n) |
threads.inuse | N | Threads in use (n) |
threads.full | n | Thread pool is full (n) |
threads.waiting | N | Threads waiting for pool (n) |
threads.fault | n | Threads wait ending on timeout (n) |
contexts.count | N | User context allocated (n) |
contexts.inuse | N | User context in use (n) |
contexts.fault | n | User context fault (n) |
workers.exec.count | n | Worker execution count (n) |
forms.exec.hit | n/s | Forms hit count (n/s) |
forms.count | N | Forms count (N) |
exceptions.va.count | n | Exceptions VA count (n) |
exceptions.hit | n/s | Exceptions hit count (n/s) |
exceptions.count | n | Exceptions count (n) |
events.hit | n/s | Events hit count (n/s) |
events.count | n | Events count (n) |
http.hit | n/s | HTTP hit count (n/s) |
http.count | n | HTTP count (n) |
http.redirect.hit | n/s | HTTP redirect hit count (n/s) |
http.redirect.count | n | HTTP redirect count (n) |
http.acl.deny.hit | n/s | HTTP ACL deny count (n/s) |
http.acl.deny.count | n | HTTP ACL deny count (n) |
http.500.hit | n/s | HTTP 500 hit count (n/s) |
http.500.count | n | HTTP 500 count (n) |
http.400.hit | n/s | HTTP 400 hit count (n/s) |
http.400.count | n | HTTP 400 count (n) |
authenticate.success.count | n | login success count (n) |
authenticate.failure.hit | n/s | login failure hit count (n/S) |
authenticate.failure.count | n | login failure count (n) |
task.execute.hit | n/s | Task execute hit (n/s) |
task.execute.count | n | Task execute count (n) |
task.running.count | N | Task running count (N) |
task.success.hit | n/s | Task success hit (n/s) |
task.success.count | n | Task success count (n) |
task.failure.hit | n/s | Task failure hit count (n/s) |
task.failure.count | n | Task failure count (n) |
task.retry.hit | n/s | Task retry hit count (n/s) |
task.retry.count | n | Task retry count (n) |
task.deadletter.hit | n/s | Task deadletter hit count (n/s) |
task.deadletter.count | n | Task deadletter count (n) |
task.deadletter.failure.hit | n/s | Task deadletter failure hit count (n/s) |
task.deadletter.failure.count | n | Task deadletter failure count (n) |
Les compteurs de performances peuvent être collectés :
- Par interrogation du serveur de performance.
- Par collecte des évènements de monitoring.
Les compteurs de performances peuvent déclencher des alertes ou des erreurs sur dépassement de seuils pré-configurés.