Skip to content

R2. Chart Metric Reference by Screen

Diátaxis: Reference · ← Back to contents

A dictionary of what each chart means, what counts as a problem, what else to look at, and what problem to expect. Pressing the ? icon beside a chart title on the screen shows the same explanation, and the AI Diagnosis button analyses the actual values on the basis of it.

Note Path = where to find it from the left menu (▸ separates menu levels) · Meaning = the value the chart shows · Problem signals and checks = the risky patterns and thresholds, plus what else to check and the problem to expect. Caution The thresholds are general guidance. Depending on the character of the service, the change against the usual baseline can matter more.

R2.1 WAS Charts​

Throughput and Response (Dashboard, Application, Statistics)​

TPS​

  • Path -- WAS ▸ Application ▸ TPS tab · WAS ▸ Dashboard widget
  • Meaning -- Transactions per second, showing the application's throughput.
  • Problem signals and checks -- Generally, higher is better, and a sudden drop means a service failure. Compare it with response time to diagnose the bottleneck.

Active Users​

  • Path -- WAS ▸ Application ▸ Active Users tab · WAS ▸ Dashboard widget
  • Meaning -- Shows the number of unique active users within a configurable time range (5 minutes by default).
  • Problem signals and checks -- Judge whether it is unusual by comparing it with the usual traffic pattern. A high value outside business hours suggests abnormal traffic or bot activity.

User Satisfaction Index​

  • Path -- WAS ▸ Application ▸ User Satisfaction tab · WAS ▸ Dashboard widget
  • Meaning -- Shows the Application Performance Index from 0 to 100.
  • Problem signals and checks -- 94 to 100: excellent, 85 to 93: good, 70 to 84: fair, 50 to 69: poor, below 50: critical. It is calculated from the proportion of Satisfied (< T), Tolerating (T to 4T), and Frustrated (> 4T or an error) responses.

Error Rate​

  • Path -- WAS ▸ Application ▸ Error Rate tab
  • Meaning -- Shows the proportion of 4xx/5xx error responses against all requests.
  • Problem signals and checks -- The top shows the error rate (%) and the bottom the error count against the request count. The ideal value is close to 0%, and when the error rate rises, check the error log, the recent deployment history, and whether a dependent service has failed.

Average Response Time​

  • Path -- WAS ▸ Application ▸ Average Response Time tab
  • Meaning -- Shows the application's average response time in milliseconds (ms).
  • Problem signals and checks -- 2 seconds or less is normal; from 3 seconds users perceive the delay, and 12 seconds or more is serious. Check slow SQL, external API calls, GC pauses, and resource contention.

Response Time​

  • Path -- WAS ▸ Application ▸ Response Time tab
  • Meaning -- Shows the maximum, minimum, and average response time together with the request count.
  • Problem signals and checks -- A large gap between the average and the maximum means some requests are abnormally slow. Check the outlier transactions in the traces.

Request Count​

  • Path -- WAS ▸ Application ▸ Request Count tab
  • Meaning -- Shows the total request count over time.
  • Problem signals and checks -- Useful for grasping the traffic volume pattern. Use it for capacity planning and for identifying traffic spikes.

Response Codes​

  • Path -- WAS ▸ Application ▸ Response Code tab
  • Meaning -- Shows the response codes classified into the 2xx/3xx/4xx/5xx categories.
  • Problem signals and checks -- In a normal state most are 2xx. A rise in 4xx indicates client errors such as a wrong URL or an authentication failure; a rise in 5xx indicates server errors such as a bug or resource exhaustion.

Delayed Transactions​

  • Path -- WAS ▸ Application ▸ Delayed Transactions tab
  • Meaning -- Shows the transactions in flight, split into Satisfying (under 3 seconds), Tolerating (3 to 12 seconds), and Frustrating (12 seconds or more).
  • Problem signals and checks -- A rising Frustrating count means many users are waiting. Investigate the thread dump and the traces of the slow transactions.

Session Count​

  • Path -- WAS ▸ Application ▸ Session tab
  • Meaning -- Shows the number of HTTP sessions created on the WAS.
  • Problem signals and checks -- A sudden rise suggests a session leak (sessions not released). An excessive session count raises memory usage, so check the session timeout and invalidation policy.

Average Database Response Time​

  • Path -- WAS ▸ Dashboard widget
  • Meaning -- Overlays the database processing time on the overall average response time.
  • Problem signals and checks -- Shows how much of the overall response time the database accounts for. Where the database share is large, consider SQL optimization, adding indexes, and query tuning.

Real-Time Requests​

Request Velocity​

  • Path -- WAS ▸ Request Velocity · WAS ▸ Dashboard top widget
  • Meaning -- Shows the requests in flight per instance as a stacked bar chart, split into the Satisfying/Tolerating/Frustrating bands.
  • Problem signals and checks -- A tall bar means many requests are in flight. When the Tolerating and Frustrating bands grow, check for slow queries, external API delays, and thread pool exhaustion.
  • Application group band -- A band above the bars tells you which group a stretch belongs to. With dozens of instances the names alone rarely reveal the business function, and where names change on every deployment (Kubernetes) they reveal nothing at all. Boundary lines separate the stretches, and hovering a bar shows its group in the tooltip.
    • The band appears only when the sort is application group (the WAS default). Under any other sort the same group scatters and the band would fragment, so it is hidden automatically.
    • Band names honour display names (aliases). Grouping and sorting still run on the original names.
    • Instances whose group could not be resolved are not dropped -- they collect into an unnamed stretch at the end.
    • The same band appears on the same widget in My Dashboard.

Request Viewer​

  • Path -- WAS ▸ Request Velocity · WAS ▸ Dashboard top widget
  • Meaning -- Visualizes the concurrent user count (the number on the left) and the flow of live requests as a missile animation.
  • Problem signals and checks -- The circular wall represents the distribution of requests by response time band. When the wall grows orange or red, investigate the cause of the slow transactions.

Topology (T-Map)​

T-Map​

  • Path -- WAS ▸ Topology Map · WAS ▸ Application ▸ Transaction Map (T-Map) tab
  • Meaning -- Shows the response time distribution over time as a heatmap.
  • Problem signals and checks -- The X axis is time and the Y axis is response time (seconds). Points dense at the bottom (fast responses) are normal; when points appear at the top (slow responses), investigate the trace detail of those transactions.

JVM​

JVM Heap Utilization​

  • Path -- WAS ▸ JVM ▸ Memory Utilization tab
  • Meaning -- Shows the JVM heap memory utilization (%).
  • Problem signals and checks -- The normal pattern is a sawtooth (falling after GC); staying above 80%, or a flat pattern, suggests a memory leak. Check the heap dump analysis, large object allocation, and the cache settings.

JVM Memory Size​

  • Path -- WAS ▸ JVM ▸ Memory Size tab
  • Meaning -- Shows the committed, used, and max memory sizes of the JVM heap and PermGen/Metaspace.
  • Problem signals and checks -- When used approaches max there is a risk of an OutOfMemoryError. Raise -Xmx or optimize memory use.

GC Time​

  • Path -- WAS ▸ JVM ▸ GC Time tab
  • Meaning -- Shows the time GC takes and the GC count.
  • Problem signals and checks -- Frequent long GC pauses of 500 ms or more make the application stall. Check the heap size, memory leaks, the object creation rate, and the GC algorithm settings.

Thread States​

  • Path -- WAS ▸ JVM ▸ Thread State tab
  • Meaning -- Shows the distribution of thread states (Runnable, Blocked, Waiting, and so on).
  • Problem signals and checks -- Many blocked threads suggest lock contention; many waiting threads suggest thread pool exhaustion. Analyse the cause with a thread dump.

Loaded Class Count​

  • Path -- WAS ▸ JVM ▸ Loaded Classes tab
  • Meaning -- Shows the number of loaded classes, the total class count, and the unloaded class count.
  • Problem signals and checks -- A class count that keeps rising with no unloading suggests a classloader leak. It is common in environments with frequent redeployment and can cause a Metaspace OutOfMemoryError.

Open File Count​

  • Path -- WAS ▸ JVM ▸ Open Files tab
  • Meaning -- Shows the number of file descriptors currently open and the maximum allowed.
  • Problem signals and checks -- Approaching the maximum produces a 'Too many open files' error. Check for socket leaks, unclosed streams, and the ulimit setting.

JVM CPU Time​

  • Path -- WAS ▸ JVM ▸ CPU Utilization tab
  • Meaning -- Shows the CPU time the JVM process consumes.
  • Problem signals and checks -- A high value indicates computation-intensive work is running. Compare it with the system-wide CPU to see whether the JVM is the main consumer.

Datasources​

Datasource Pool Count​

  • Path -- WAS ▸ Datasources
  • Meaning -- Shows the number of database connections in use from the connection pool.
  • Problem signals and checks -- Approaching the maximum pool size means database connection exhaustion. Check slow queries, connections not being returned, and the pool size setting.

Containers​

CPU Utilization​

  • Path -- WAS ▸ Server/Container
  • Meaning -- Shows the CPU utilization of all Docker containers on the host.
  • Problem signals and checks -- Monitors how CPU resources are distributed between containers at host level. Identify a container taking excessive CPU.

CPU Usage​

  • Path -- WAS ▸ Server/Container
  • Meaning -- Shows the CPU usage of all Docker containers on the host in absolute terms.
  • Problem signals and checks -- Lets you compare CPU consumption per container. Check the usage against the host's total CPU capacity.

Memory Usage​

  • Path -- WAS ▸ Server/Container
  • Meaning -- Shows the memory usage of all Docker containers on the host.
  • Problem signals and checks -- When the total container memory approaches the host memory there is a risk of an OOMKill. Compare the per-container memory limits with the actual usage.

Disk Usage​

  • Path -- WAS ▸ Server/Container
  • Meaning -- Shows the disk I/O throughput of all Docker containers on the host.
  • Problem signals and checks -- Lets you see which container generates excessive disk I/O. Check log files, temporary file creation, and so on.

Traffic​

  • Path -- WAS ▸ Server/Container
  • Meaning -- Shows the network bytes sent and received by all Docker containers on the host.
  • Problem signals and checks -- Monitors traffic between containers and external traffic. When an abnormal traffic pattern is detected, check that container.

Packet Errors​

  • Path -- WAS ▸ Server/Container
  • Meaning -- Shows the network error count of all Docker containers on the host.
  • Problem signals and checks -- In a normal state it should be 0. When errors occur, check the network configuration, the Docker network driver, and the connections between containers.

Container Status​

  • Path -- WAS ▸ Server/Container
  • Meaning -- Shows the overall resource usage of the containers -- CPU, memory, network, and so on -- at a glance.
  • Problem signals and checks -- Useful for grasping the state of every container quickly. Identify a container with a problem and analyse it in detail in the individual charts.

R2.2 Web Server Charts​

Traffic​

  • Path -- Web Server ▸ Web Performance · Web Server ▸ Dashboard
  • Meaning -- Shows the bytes per second the web server handles, monitoring bandwidth use.
  • Problem signals and checks -- A sudden surge suggests a DDoS attack or a large file download. Compare it with the usual pattern to identify unusual traffic.

Requests per Second​

  • Path -- Web Server ▸ Web Performance · Web Server ▸ Dashboard
  • Meaning -- Shows the number of requests the web server handles per second.
  • Problem signals and checks -- The higher the value, the heavier the load. Compare it with the WAS TPS to see whether the web server is the bottleneck.

Worker States​

  • Path -- Web Server ▸ Web Performance · Web Server ▸ Dashboard
  • Meaning -- Shows the number of active and idle workers.
  • Problem signals and checks -- When every worker is busy, no new request can be handled. Raise MaxWorkers or optimize the backend processing speed.

Connection States​

  • Path -- Web Server ▸ Web Performance · Web Server ▸ Dashboard
  • Meaning -- Shows the connection states (Reading, Writing, Keep-alive, Idle) item by item.
  • Problem signals and checks -- Many in Reading suggest slow clients; many in Writing suggest a slow backend. The overall connection distribution reveals the type of load on the server.

R2.3 DBMS Charts​

The metric charts are viewed on the metric tab of Database ▸ Cluster View / Instance View, and the diagnosis and SQL analysis charts under Database ▸ DBMS Diagnosis. The same description card appears above each chart.

Connection​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the number of clients currently connected and the maximum connections allowed.
  • Problem signals and checks -- When utilization passes 80% of max_connections, new connections can be refused, so raising max_connections or checking the connection pool is needed.

QPS​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- The number of SELECT/INSERT/UPDATE/DELETE and other queries the server has run in the last few seconds.
  • Problem signals and checks -- A sharp rise against usual can mean a traffic peak or abnormal polling or retries, so check it together with the slow queries.

DML Ratio​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the execution ratio of Select / Insert / Update / Delete, stacked.
  • Problem signals and checks -- An abnormally high Update/Delete share can cause lock contention and binlog bloat.

Thread Running​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the number of threads actually doing work (using CPU or IO).
  • Problem signals and checks -- Continuously exceeding the vCPU count raises the cost of context switching and delays responses.

Session Analytics​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the sessions in progress, the commands running, and the wait states at a glance.
  • Problem signals and checks -- If many sessions have a long TIME and a STATE of 'Locked' or 'Waiting for ...', check for lock contention or slow queries.

Buffer Pool Allocation Ratio​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the proportion of memory allocated to the InnoDB buffer pool and its share of OS memory.
  • Problem signals and checks -- Taking 70 to 80% or more of OS memory can cause swapping and poor IO performance from a lack of OS cache.

Buffer Pool Hit Ratio​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the hit ratio -- the proportion where data could not be read from the buffer pool and disk IO occurred.
  • Problem signals and checks -- A hit ratio falling below 95% means disk reads have become excessive, so consider raising innodb_buffer_pool_size.

Buffer Pool Page Used Ratio​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the distribution of data pages (used) and empty pages (free) in the buffer pool.
  • Problem signals and checks -- With almost no free pages, caching new data becomes difficult, so the buffer pool size or LRU tuning needs attention.

IO Reads/Writes​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the number of InnoDB disk read, write, and fsync calls.
  • Problem signals and checks -- A surge in reads against usual points to queries not using indexes; a surge in writes or fsync points to possible bulk transactions.

IO Reads/Writes Pending​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- The number of InnoDB read, write, and fsync requests waiting unprocessed.
  • Problem signals and checks -- A pending value continuously above 0 means disk IO is the bottleneck, so the storage IOPS or io_capacity needs checking.

Waits​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the row lock, log, and buffer pool wait counts, stacked.
  • Problem signals and checks -- A surge in row lock waits means lock contention; a surge in log waits means a log buffer or disk synchronization delay.

InnoDB Engine Status​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Analyses the SHOW ENGINE INNODB STATUS output to show the transaction, deadlock, and IO state.
  • Problem signals and checks -- When the LATEST DETECTED DEADLOCK entry updates often, the transaction isolation level or the index design needs checking.

InnoDB Engine Mutex​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the OS wait count per internal InnoDB mutex.
  • Problem signals and checks -- A continuous rise in OS waits on a particular mutex is a concurrency bottleneck, so consider separating the workload or upgrading the version.

Table Open Caches​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the efficiency of table metadata caching as the hit ratio of table_open_cache.
  • Problem signals and checks -- A low hit ratio with opened tables rising quickly means table_open_cache has to be raised.

Thread Cache​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the efficiency of thread reuse as the thread_cache hit ratio.
  • Problem signals and checks -- A low hit ratio means a new thread is created for every connection, raising the load, so consider raising thread_cache_size.

Thread Pool​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the active and idle thread counts of the thread pool plugin and the wait queue size.
  • Problem signals and checks -- A queue that keeps filling suggests too small a thread_pool_size, or congestion from slow queries.

Binary Log Cache​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the use of the cache where transaction events are held before being written to the binlog file.
  • Problem signals and checks -- Frequent disk use means binlog_cache_size is too small, so split large transactions or raise the cache size.

Query Cache​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the hit ratio and memory use of the query cache (removed from MySQL 8.0).
  • Problem signals and checks -- On a write-heavy system the cost of cache invalidation is high, so disabling query_cache is recommended.

Open Tables​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the number of tables the server currently has open.
  • Problem signals and checks -- Approaching the table_open_cache limit makes open and close repeat, raising IO.

Opened Tables​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the cumulative table open count since the server started.
  • Problem signals and checks -- A large rise in a short time signals that table_open_cache is too small and cache eviction happens often.

Lock Waits​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows how many times a table lock was acquired immediately and how many times it had to wait.
  • Problem signals and checks -- A rising waited ratio means lock contention is likely, from long transactions or missing indexes.

Row Lock​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the InnoDB row lock wait count and the average wait time.
  • Problem signals and checks -- A long average wait or a rising wait count points to a hotspot row or a missing index.

Meta Data Lock​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the sessions waiting or blocked by a metadata lock.
  • Problem signals and checks -- A user transaction held open for a long time during DDL can stop all operational work.

Lock Contention​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows lock contention statistics per table and row.
  • Problem signals and checks -- Lock contention concentrated on one table calls for index redesign or splitting the transaction.

Fullscan Ratio​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the proportion of all queries where a full scan occurred.
  • Problem signals and checks -- The higher the ratio, the more likely an index is missing, so analyse it together with the slow query log.

FullScan Ratio per User​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the full scan rate per user (account).
  • Problem signals and checks -- If full scans are frequent for only one user, check that application's queries first.

Tmp Table Used Ratio​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the temporary table creation count and ratio.
  • Problem signals and checks -- When temporary tables are created often through GROUP BY, ORDER BY, and the like, consider indexes or rewriting the query.

Tmp Table Disk Used Ratio​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the proportion of in-memory temporary tables written to disk.
  • Problem signals and checks -- A high disk use ratio means tmp_table_size / max_heap_table_size is too small, or the BLOB/TEXT usage pattern needs checking.

DataSize per Storage Engine​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the data size distribution per storage engine (InnoDB, MyISAM, and so on).
  • Problem signals and checks -- A large share on an engine other than InnoDB calls for considering migration, for transaction and recovery stability.

DataSize per Schema​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the data and index size distribution per schema.
  • Problem signals and checks -- An abnormally large schema calls for checking the retention policy (partitioning, archiving).

Tables per Schema/Engines​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the table count distribution per schema and storage engine.
  • Problem signals and checks -- MyISAM and InnoDB mixed in one schema can cause problems with backup and recovery consistency and with transaction handling.

Tables without PK​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the tables with no PRIMARY KEY defined.
  • Problem signals and checks -- Without a PK, InnoDB uses a 6-byte hidden ROWID, which harms replication stability and performance.

Table without Index​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the tables with no index at all.
  • Problem signals and checks -- With no index every query is a full scan, so it slows sharply as the data grows.

Table with Non-Numeric PK​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the tables whose PK is not numeric.
  • Problem signals and checks -- A string PK enlarges the InnoDB clustered index and raises IO cost, so BIGINT/INT is recommended where possible.

Table IndexSize > DataSize​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the data-to-index size ratio per table.
  • Problem signals and checks -- Where the index is larger than the data, check for unnecessary or duplicate indexes.

Server Info​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the basic server information -- the DBMS version, host name, OS information, and so on.

Uptime​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows how long the server has been running since its last restart.
  • Problem signals and checks -- If it is unexpectedly short, check for an abnormal termination or a restart by the OOM killer.

Server Vars&State​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the server's main system variables and status variables at a glance.

Logical Data Size​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the logical data size of all schemas.
  • Problem signals and checks -- A sharp rise can be the effect of bulk inserts or an added index, so check the free disk space alongside it.

CPU Detail​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows CPU utilization (user/sys/wait/idle).
  • Problem signals and checks -- User CPU staying above 80% points to query tuning; high IO wait points to a disk bottleneck.

Memory​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows total memory usage and the InnoDB buffer pool share.
  • Problem signals and checks -- Almost no free memory with rising swap use means a memory shortage, so the buffer pool or OS memory needs checking.

Load Average​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the 1-, 5-, and 15-minute system load averages.
  • Problem signals and checks -- Staying above the vCPU count means processing capacity has reached its limit.

Disks​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the usage and read/write IO per disk.
  • Problem signals and checks -- Utilization of 90% or more can make InnoDB unable to operate, so it needs immediate action.

Network​

  • Path -- Database ▸ Instance View (metric tab -- the same in Cluster View)
  • Meaning -- Shows the traffic sent and received per network interface.
  • Problem signals and checks -- Continuous link saturation can cause replication lag or client timeouts.

Parameter Diagnosis​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Compares the recommended and current values of the main parameters to diagnose problem items.
  • Problem signals and checks -- Items marked critical or warning affect operational stability, so consider applying them first.

Characterset/Collation​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the character set and collation distribution per table and column.
  • Problem signals and checks -- Mixed collations can prevent an index being used in a JOIN.

Most Latency Indexes​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the index-using SQL with the greatest latency.
  • Problem signals and checks -- When one index keeps appearing, the index shape or the query pattern needs redesigning.

Duplicate Indexes​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the indexes duplicated on the same column combination.
  • Problem signals and checks -- Duplicate indexes only raise INSERT/UPDATE cost, so clearing them improves write performance.

Unused Indexes​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the indexes with no recent usage.
  • Problem signals and checks -- An index unused for a long time is a candidate for removal. Check first whether it is used only at a particular time (month end, for example) before deleting it.

TOP SQL 50​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the top 50 SQL by execution frequency and cumulative time.
  • Problem signals and checks -- When the top 1 to 5 account for most of the load, tuning those first is the most effective.

Disk Tmp Use 50​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the 50 SQL that used disk temporary tables the most.
  • Problem signals and checks -- Frequently exceeding the in-memory temporary table limit calls for considering tmp_table_size or index and query tuning.

Slow Query TOP 50​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the top 50 slow queries that passed the given threshold time.
  • Problem signals and checks -- When the same query keeps appearing, improving that one alone improves overall performance considerably.

Fullscan TOP 50​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the top 50 SQL where full scans occurred most.
  • Problem signals and checks -- The cost of a full scan grows linearly with the data, so consider adding an index first.

CPU Usage TOP 50​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the top 50 SQL by CPU usage.
  • Problem signals and checks -- Where complex functions or expressions, or sorting and aggregation, are the cause, rewriting the query can improve it greatly.

Slow Query​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the full list of slow queries that passed the threshold time.
  • Problem signals and checks -- When a pattern of concentration at a particular time appears, check whether it clashes with batch or scheduled work.

Memory Stat. by Accounts​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the cumulative memory usage per user account.
  • Problem signals and checks -- An abnormally large figure for one account can mean a prepared statement or result set leak in the application.

Memory Stat. by Hosts​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the cumulative memory usage per connecting host.
  • Problem signals and checks -- Large usage from an unexpected host calls for confirming where it comes from and checking the network policy.

Memory Stat. by Threads​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the memory usage per thread.
  • Problem signals and checks -- When one thread holds excessive memory, check that session's query and temporary table usage pattern.

Memory Stat. by Events​

  • Path -- Database ▸ DBMS Diagnosis
  • Meaning -- Shows the memory usage per performance_schema event.
  • Problem signals and checks -- When one event grows abnormally, check for a memory leak in the performance counters.

CAS Usage​

  • Path -- Database ▸ Instance View (CUBRID)
  • Meaning -- The CAS (Common Application Server) slot utilization. It shows the share of application requests the broker is handling at once.
  • Problem signals and checks -- Continuously high utilization calls for considering more CAS (MAX_NUM_APPL_SERVER) or tuning the connection pool.

Job Queue​

  • Path -- Database ▸ Instance View (CUBRID)
  • Meaning -- The number of requests waiting in the broker job queue.
  • Problem signals and checks -- A queue that keeps building suggests too few CAS, or delay from slow queries.

TPS/QPS​

  • Path -- Database ▸ Instance View (CUBRID)
  • Meaning -- Transactions per second (TPS) and queries per second (QPS).
  • Problem signals and checks -- On a sharp change against usual, check for a change in the workload or abnormal traffic.

Long Trans​

  • Path -- Database ▸ Instance View (CUBRID)
  • Meaning -- The number of long-running transactions.
  • Problem signals and checks -- Long transactions increase lock holding and MVCC burden, so check the query behind them.

Error Query​

  • Path -- Database ▸ Instance View (CUBRID)
  • Meaning -- The number of failed (error) queries.
  • Problem signals and checks -- When error queries rise, check the broker log and the application queries together.

R2.4 System (Server) Charts​

CPU, Memory, and Load​

CPU Utilization​

  • Path -- Server ▸ Resources
  • Meaning -- Shows the system CPU utilization by category -- user, system, IOWait, and so on.
  • Problem signals and checks -- High IOWait indicates a disk I/O bottleneck; a high system% indicates kernel overhead. Diagnose the cause of the slowdown from the proportions per category.

Detailed CPU Utilization​

  • Path -- Server ▸ Resources
  • Meaning -- Shows every CPU utilization item in detail -- user, system, IOWait, steal, nice, SoftIRQ, and so on.
  • Problem signals and checks -- Useful for diagnosing virtualization overhead (steal) or an interrupt storm (SoftIRQ). When one item is abnormally high, focus the checks on that area.

Memory Utilization​

  • Path -- Server ▸ Resources
  • Meaning -- Shows the system physical memory and swap usage.
  • Problem signals and checks -- High swap use means physical memory is short and performance degrades seriously. Add RAM or reduce the memory-consuming processes.

Load Average​

  • Path -- Server ▸ Resources
  • Meaning -- Shows the system load average at 1-, 5-, and 15-minute intervals.
  • Problem signals and checks -- Exceeding the CPU core count is an overload. When the 1-minute value is far above the 15-minute value, the load is rising, so find the cause.

Disks​

Disk Usage​

  • Path -- Server ▸ Disks
  • Meaning -- Shows disk usage per partition.
  • Problem signals and checks -- Passing 90% is risky and can cause a service failure. Monitor the rate of growth and plan capacity.

I/O Count​

  • Path -- Server ▸ Disks
  • Meaning -- Shows the disk read/write IOPS (I/O operations per second).
  • Problem signals and checks -- High IOPS together with rising response time means the disk is saturated. Consider replacing it with an SSD or optimizing I/O.

I/O Throughput​

  • Path -- Server ▸ Disks
  • Meaning -- Shows the disk read/write byte throughput.
  • Problem signals and checks -- Shows the actual data transfer rate. Compare it with the disk specification to see whether it is saturated.

Average Service Time​

  • Path -- Server ▸ Disks
  • Meaning -- Shows the average service time of I/O requests.
  • Problem signals and checks -- Passing 20 ms means the disk is slow. Check the disk state, the RAID configuration, and the I/O scheduler setting.

Average Queue Length​

  • Path -- Server ▸ Disks
  • Meaning -- Shows the average length of the I/O request queue.
  • Problem signals and checks -- Passing 2 to 3 means the disk cannot keep up with the requests. Requests wait in the queue, so consider faster disks or spreading the I/O.

File Node Count​

  • Path -- Server ▸ Disks
  • Meaning -- Shows the inode usage of the file system.
  • Problem signals and checks -- Once inodes are exhausted, no new file can be created even with free disk space. It mainly happens where there are a great many small files, so clear the unnecessary ones.

Network​

TCP Connection States​

  • Path -- Server ▸ Network
  • Meaning -- Shows the TCP connection states (ESTABLISHED, TIME_WAIT, CLOSE_WAIT, and so on) item by item.
  • Problem signals and checks -- Many in CLOSE_WAIT indicate a connection leak; many in TIME_WAIT after heavy traffic indicate possible port exhaustion.

Traffic​

  • Path -- Server ▸ Network
  • Meaning -- Shows the bytes sent and received per second per network interface.
  • Problem signals and checks -- Monitors bandwidth utilization. Passing 70% of the interface capacity calls for considering network expansion.

Transmitted Packets​

  • Path -- Server ▸ Network
  • Meaning -- Shows the packets sent and received per second per interface.
  • Problem signals and checks -- Many packets with few bytes means small-packet overhead (possibly an attack). Work out the average packet size (bytes/packets) to judge whether it is unusual.

Packet Errors​

  • Path -- Server ▸ Network
  • Meaning -- Shows the network error count per interface.
  • Problem signals and checks -- Errors mean a hardware problem, a faulty cable, or a driver bug. In a normal state it should be 0.

Source: question.json in the product (the chart ? tooltips) and chart-descriptions.js (the DBMS description cards). The ? icon on the product screen is the single source of truth for the current thresholds and descriptions.