Skip to main content

Multiple Instance Monitoring

Home > Select Project > Dashboard > Monitoring Multiple Instances

While you can monitor only one instance in Monitoring a Database Instance, Monitoring Multiple Instances allows you to see and compare the metrics for multiple instances on a screen. You can quickly see the status of instances through grouped values of instance metrics such as running agent count, total number of active sessions, and transaction information. Graph charts make it easy to see how instances' metric values are trending and which instances have higher metric values. It provides information of the longest lasting active session.

Multi-instance Monitoring

Tip

If you select an agent to compare in Dashboard > Instance List, and then select Multi view, the Monitoring Multiple Instances menu appears.

Viewing the past data

The dashboard basically provides real-time monitoring features in which you can view the past data.

Real-time modeNon-real-time mode
Real-time modeNon-real-time mode
  1. Select Pause icon in the time selector at the upper left of the screen.

  2. Click the date and time to set the desired time zone.

    Configure manual time

  3. Select Apply.

It updates the data in widgets placed on the dashboard based on the set time. To switch to the Realtime mode, select Playback icon.

Checking the agent

Checking the agent connection status

On the upper left of the screen, the right area of the time selector provides information that allows you to check the status of agents connected to the project. This allows you to immediately check whether the target application server is running.

  • Total: Number of all agents connected to the project

  • Active: Number of active agents

  • Inactive: Number of inactive agents

  • Agent display icon: Can display or hide inactive agents.

Agent-based monitoring

Selecting the agent

By default, the dashboard displays the metrics collected from all agents in charts and you can also view the data by agent. Select one or more agents under the time selector. The data of the widgets on the dashboard is updated with the metrics of the selected agent(s).

Tip

To select all agents again while one or more agents is selected, deselect them or select Total.

Note

If there are numerous agents connected to the project, it is efficient to set the agent names to be short. For more information about the agent name setting, see the following.

Monitoring each classified agents

Select agent by category

You can select and monitor agents by each group classified through agent settings.

  • Agent: You can select an individual agent or all agents.

  • Agent by cluster: In case of the database that consists of multiple clusters, agents can be monitored by cluster.

Editing a dashboard widget

The widgets on the dashboard can be adjusted to the desired size and placed in the desired positions. You can also delete unnecessary widgets or add them again.

Resizing a widget

Resizing a widget

Click and hold the mouse on the Resize icon element at the lower right of the widget, and then drag it to the desired size. Grids with a uniform horizontal-to-vertical ratio appear and the size of the widget can be adjusted in each grid.

Moving a widget

Moving a widget

When you move the mouse cursor to the top of the widget, the cursor shape changes to Move icon. At this time, you can move the widget by dragging with your left mouse button and move to the desired position.

Deleting a widget

Deleting a widget

Right-click on a widget to delete. If you select Delete, the widget is deleted from the dashboard.

Adding a widget

Adding widget

Move the mouse cursor to an empty space on the dashboard and then right-click on it. Select a widget to add from the pop-up menu. Place the widget to the desired position and then resize it.

Note
  • For more information about the widgets that can be placed on the dashboard, see the following.

  • The widgets that can be added are fixed, but we plan to increase supports through future updates.

Using the widget options

The icons displayed on the widget function as follows:

  • Information icon: You can see the features and information for major widgets. (multilingual support planned)

  • Up arrow icon or Down arrow icon: You can enlarge or reduce the range of the vertical axis.

  • Fullscreen icon: You can view the widget's data on a wider screen.

  • Detail icon : A modal window appears where you can search the data of the widgets separately by agent.

Note

The provided options may differ depending on the widget.

Preset

Preset

You can save and load the settings and layouts for widgets on the dashboard. The default preset (Default) cannot be changed, but you can create a new preset by resizing and placing widgets in the desired format.

Creating a new preset

  1. Place the widgets in the desired format on the dashboard. You can also resize and place only the widgets to be frequently checked.

  2. On the upper right of the screen, select Save icon.

  3. Enter a new preset name.

    Save Preset

    To save the selected agent data, select Include agent selection history.

  4. Select Save.

You can see the newly saved preset from the preset list.

Note
  • If any changes are made to the newly created preset, save the preset again. Select Save icon and then save the preset with the same name. Any changes are overwritten in the existing preset.

  • If you go to another menu without saving changes on the dashboard, the changes are not saved.

  • Presets are saved in your user account and cannot be shared with other users. We plan to update it for better features.

Deleting presets

If you have unused presets, you can delete them from the preset list. Select Delete icon on the right of the item to delete from the preset list.

Learning about the main widgets

The types of widgets that can be placed on the dashboard are as follows.

  • DB Status

    You can see the status of agents (instances) in the project. Individual agents are displayed in a hexagon. If you hover your mouse over it, the current status and IP address appear.

    DB Status

    The status of the agent can be classified as follows:

    • Normal: Number of normal agents.

    • Inactive Agents: Number of disconnected agents.

  • [XOS] CPU

    You can see the CPU usages for each agent in real time.

  • [XOS] Memory

    You can see the memory usages for each agent in real time.

  • XOS Disk Usage

    It displays the disk utilization, free sizes, and total sizes for each agent. If you select Icon on the upper right of the widget, you can see them on a wider screen.

    Tip

    If there is no data displayed, add the following settings to the xos.conf file.

    xos.conf
    # Off if the disk usage is 0; unit: second
    disk_usage_interval=60
  • Read requests

    A widget that measures the number of read requests per second. Read requests are tasks involving queries, aggregation, index scans, and the like that query or retrieve data from the database. It is useful to detect any changes in database usage patterns.

    Calculation: query$opcounters + getmore$opcounters

  • Write requests

    A widget that measures the number of write requests per second. Write requests are tasks involving addition, update, and deletion in the database. It is useful to detect any changes in database usage patterns.

    Calculation: insert$opcounters + update$opcounters + delete$opcounters

  • Read Latency (latency$reads$opLatencies)

    A widget that measures the latency (in microseconds) of read tasks that is the time taken for a client to request a read task and perform the task in the database. If a read task has a high latency consider the way to improve the performance by optimizing indexes or improving queries.

  • Write Latency (latency$writes$opLatencies)

    A widget that measures the latency (in microseconds) of write tasks that is the time taken for a client to request a write task and perform the task in the database. If a write task has a high latency consider the way to improve the performance by optimizing indexes or performing a batch job for writing.

  • Page Faults (page_faults$extra_info)

    This widget measures the number of page faults that occur. A page fault occurs while in memory management and occurs when a memory page requested by a process does not exist in physical memory and the page must be read from the disk. At this time, disk I/O may occur and the task may be delayed. This usually means that the system is running out of memory.

    You can check whether the read and write tasks in the database depend on the disk I/O and whether low memory caused performance issues.

    Note

    For more information about the page faults, see the following link.

  • Active Connections (active$connections)

    A widget that measures the number of connections by active clients. You can monitor the number of connections from the clients that the database server can handle at the same time. A high number of connections means you need to increase server capacity. In the opposite case, a problem may have occurred. Accordingly, it is better to quickly check the system status.

  • Read Clients (readers$activeClients$globalLock)

    A widget that measures the number of clients performing read tasks in the MongoDB's global lock state. You can see how many clients the database server handles for read tasks. This is a useful metric for monitoring the database performance and tracing response times. Excessive read tasks may cause performance degradation.

  • Write Clients (writers$activeClients$globalLock)

    This widget measures the number of clients performing write tasks in the MongoDB's global lock state. You can see how many clients the database server handle for write tasks. This metric is useful for monitoring the database performance and tracing response times. Excessive write tasks may cause performance degradation.

  • Queued Read Requests (readers$currentQueue$globalLock)

    This widget measures the number of clients waiting for a global lock to process pending read tasks in the current system. It provides the data to evaluate the performance and scalability of the database. As pending read requests are increasing, the response time for requests from clients may increase and the throughput may reach the limit.

    If pending read tasks increase rapidly, you can consider increasing the server resources, analyze the cause of the issue, and resolve bottlenecks through measures such as query optimization and index tuning.

  • Queued Write Requests (writers$currentQueue$globalLock)

    This widget measures the number of clients waiting for a global lock to process pending write tasks in the current system. It is required to evaluate the DB performance and optimize the database. As pending write requests are increasing, the system performance may decrease or the response time for requests may increase.

    You may need to consider adding resources or identifying which tasks are causing bottlenecks.

  • Resident Memory (resident$mem)

    It is the memory size (MB) used by processes on the current system. You can see how much memory they use. Tracing memory usage optimizes the memory usage to increase the performance and becomes an important metric for efficient use of memory resources. It also helps you determine how much memory you need or whether or not to add more shards.

  • Virtual Memory (virtual$mem)

    It is the virtual memory size (MB) used by processes on the current system. You can see how much memory resources is required on the current environment. If you know the size of the virtual memory, it helps you determine how much memory is required and what measures need to be taken to reduce the virtual memory usage.

  • Network Out (bytesOut$network)

    Total number of bytes sent by the server via connections from clients or other instances. This indicates the total amount of data transmitted when the database server receives any outbound requests and returns responses. You can identify network traffic occurring on the database server, which allows you to optimize the network bandwidth or server resource usage. High network traffic is also associated with performance issues such as increase in response time.

  • Network In (bytesIn$network)

    Total number of bytes received by the server via connections from clients or other instances. It indicates the total amount of requests and data received by the database server from outside, and includes the amount of data transmitted when clients send requests and receive responses from the database server. You can see how specific applications or tasks use communication with the database. Additionally, you can plan the system capacity and evaluate scalability by understanding the trend in network traffic.

Note

A global lock is a mechanism used by MongoDB to maintain data consistency. It is applied to both read and write tasks, and acquiring a global lock is required for clients to perform the tasks.

Session Table widget

You can view real-time active sessions in the Session Table widget at the bottom of the screen.

Note

In the table list, the text colors are changed black → orangeRed, which means that the performance of the session is getting slower.

Filtering the table data

Filter

  1. Select Filter icon on the upper right of the table.

  2. Select a column header and a condition in the table.

  3. Enter a desired value in the Enter conditions field.

  4. Select Save.

Configuring the table columns

You can hide the table header columns or add any of them. You can also change the column order. Select Column icon.

Column Settings

Note
  • After configuration, select OK to apply the settings in the table.

  • In the number 3 search bar, enter text to search the desired columns. Only the columns that meet the entered text are displayed.

Adding columns

From the Number 1 list, select the items to add as table header columns. To select all items, select Select All.

Deleting columns

From the number 1 list, unselect the columns to delete. Alternatively, select Delete icon on the right of the item to delete from the number 2 list.

Changing the column order

Drag an item to reposition from the number 2 list, and then move it to the desired position.

Initializing the configuration

To cancel all changes and reset them, select Reset icon Reset.

Column information guide

The following lists the columns in the table displayed for the Active sessions section. The data is collected using the currentOp.

ItemDescription
opidUnique identifier of the operation
typeIf the value is op, the operation type can be viewed in the op field.
hostServer name
descDescription of the client including connectionId
activeIf the operation starts, the value is true and if it is idle, the value is false.
currentOpTimeStart time of the operation
opoperation type (none, update, insert, query, command, getmore, remove, etc.)
nsThe name of the namespace appears with Database or Collection.
numYieldsNumber of executed operations
waitingForLockIf the operation is waiting for lock, the value is true and if it has acquired lock, the value is false.
waitingForFlowControlIf the operation is waiting for flowcontrol, the value is true.

Downloading the view list

You can download the current Session Table list in CSV format. Select Download icon on the upper right of the Session Table section.

The naming format of the downloaded file is as follows:

  • Active sessions: activeSession_HH_MM_SS.csv

  • Lock tree: lockTree_HH_MM_SS.csv

Viewing in a new window

You can see the Session Table list on a new wider screen. Select New window icon on the upper right of the Session Table section.