How do I temporarily stop all traffic to a server; can I set Weight to 0?

Short answer

Weight cannot be set to 0; Weight is in the 1–256 range and Connection Limit in the 1–65535 range, and even at the lowest values the server continues to receive traffic. On HTTP and L7 TCP vServices, the recommended way to temporarily take a server out of traffic is Maintenance Mode: the server stops receiving new traffic without being deleted, and open connections are not cut. Depending on the situation, removing the Backend Service from the Default Backend Services list of the vService (also works on a Network vService), removing the group row, making it a Backup Service, making the health check fail on the server side or enabling maintenance mode from WEB-CLI are also used.

Problem: Taking a Server Out of Traffic Without Deleting It

When patching, updating or hardware work is to be done on a server, traffic must be moved away from it without affecting users. The first idea is to set Weight to 0; however, in TR7 Weight is at least 1. Deleting the server, on the other hand, loses its settings and vService connections, and it must be set up again when the work is finished.

Solution: Six Methods

The methods are alternatives to each other; only one is applied at a time.

Your situation	Suitable method
I want to put a single Backend Service or an entire Backend Group into maintenance	Method 1: Maintenance Mode
I want to remove the server from only one vService, or the vService is of the Network type	Method 2: Removing it from the vService list
I want to remove a single server in the group	Method 3: Removing the group row
The server should receive traffic only in an emergency	Method 4: Making it a Backup Service
I want to remove it from the server side without touching TR7	Method 5: Making the health check fail
I want to put one or many Backend Services into maintenance from the command line or with a script	Method 6: Maintenance Mode with WEB-CLI

Method 1: Maintenance Mode Recommended

• How it works The Backend Service receives no new traffic on any of the HTTP and L7 TCP vServices in which it is used; open connections continue until they end. The settings and vService connections stay in place. In a Backend Group, the mode is applied to all the servers of the group together.
• Where it is openedLoad Balancer → Backend Services (or Backend Groups) → Edit → Maintenance Mode
• Requirement Another Backend Service or a backup in the list to take over the traffic; otherwise clients receive 503
• vService type HTTP and L7 TCP (on a Network vService, Method 2 is used)

Step-by-Step Configuration

1

Adım 1 Check the vServices that will be affected

In the Used in tab of the preview panel that opens when the Backend Service is clicked, it is checked which vServices will be affected.

2

Adım 2 Open the edit form

On the Load Balancer → Backend Services screen, the Backend Service is selected and Edit is clicked (right-click → Edit can also be used). For the entire group, the group is opened on the Load Balancer → Backend Groups screen.

3

Adım 3 Turn on maintenance mode

The Maintenance Mode switch is turned on (its default value is off).

4

Adım 4 Save

The Edit button at the bottom of the form is clicked. The vServices that use the Backend Service are reloaded; ongoing connections are not cut, and new connections go to the other Backend Services. While Clustering is on and Manual Sync Mode is off (default), the setting is transferred to the peer device automatically. When the work is finished, the switch is turned off and saved in the same way; if the health check result is OK, the server receives traffic again.

Verification

1Maintenance statusTR7 side

In the list, an amber ambulance icon appears next to the name of the Backend Service, and the amber Maintenance status appears on the Live Stats screen of the vService; the current session count drops to zero as the open connections end.

ExpectedMaintenance · no new sessions

Method 2: Removing It from the vService List

• How it works The Backend Service is removed only from the list of that vService; it continues to receive traffic on other vServices. The Backend Service definition is kept.
• Where it is openedLoad Balancer → Load Balancer vService List → vService → Edit → Default Backend Services (and the Conditional Backend Services sets, if any)
• Requirement At least one other Backend Service remaining in the list; otherwise clients receive 503
• vService type HTTP, L7 TCP and Network

Step-by-Step Configuration

1

Adım 1 Check the usage

In the Used in tab of the preview panel of the Backend Service, it is checked in which vServices it is used.

2

Adım 2 Open the vService edit form

On the Load Balancer → Load Balancer vService List screen, the vService is selected and Edit is clicked.

3

Adım 3 Remove it from the list

The Backend Service is removed from the Default Backend Services list and, if any, from the lists of the conditional sets.

4

Adım 4 Save

Save is clicked. When the maintenance is finished, the Backend Service is added back to the same lists and the vService is again saved with Save.

Verification

1UsagePreview panel

That vService no longer appears in the Used in tab in the preview of the Backend Service.

ExpectedvService not in the Used in list

Method 3: Removing the Group Row

• How it works The server is removed from the Backends table of the Backend Group; the other servers of the group continue to receive traffic.
• Where it is openedLoad Balancer → Backend Groups → Edit → Backends
• Requirement At least one other row remaining in the group; when a single row remains, the delete control does not appear
• vService type HTTP, L7 TCP and Network

Step-by-Step Configuration

1

Adım 1 Open the group edit form

On the Load Balancer → Backend Groups screen, the group is selected and Edit is clicked.

2

Adım 2 Note down the row values

The address, Cookie Content, Connection Limit and Weight values in the row are noted down; when the row is removed, this information is lost.

3

Adım 3 Remove the row

In the Backends table, the row is removed with its delete control.

4

Adım 4 Save

The Edit button at the bottom of the form is clicked. When the maintenance is finished, the row is added again with the same values and the group is saved again.

Verification

1Group serversTR7 side

In the Backend Groups list, the Show Backend Services switch is turned on.

ExpectedServer not among the group rows

Method 4: Making It a Backup Service

• How it works The Backend Service turns into a backup; it does not receive traffic unless all the main Backend Services in the list are unavailable. Since it receives traffic if the main servers go down, it is not a definite "no traffic at all" solution.
• Where it is openedLoad Balancer → Backend Services → Edit → Backup Service
• Requirement Other healthy Backend Services in the list
• vService type HTTP and L7 TCP

Step-by-Step Configuration

1

Adım 1 Open the edit form

On the Load Balancer → Backend Services screen, the Backend Service is selected and Edit is clicked.

2

Adım 2 Mark it as a backup

The Backup Service switch is turned on (its default value is off).

3

Adım 3 Save

The Edit button at the bottom of the form is clicked. If the server is to return to normal distribution, the switch is turned off and saved.

Verification

1Backup statusTR7 side

In the list, an amber backup server icon appears in the name column; in Live Stats, no new sessions come to the server.

ExpectedBackup icon · no new sessions

Method 5: Making the Health Check Fail

• How it works On the server side, the endpoint polled by the health check (e.g. /health) is set to return an error; TR7 considers the server Error and takes it out of traffic. No change is required in TR7.
• Where it is opened On the Backend Service (server) side
• Requirement A Health Checks profile assigned to the Backend Service
• vService type HTTP and L7 TCP; on a Network vService, only while Port Range is off and Protocol is not ANY - L3

Step-by-Step Configuration

1

Adım 1 Check the profile

It is checked that a profile is selected in the Health Check field of the Backend Service; if it is No Check, this method does not work. If a profile is selected, step 4 is next.

2

Adım 2 Assign a profile (if needed)

On the Load Balancer → Backend Services screen, the Backend Service is selected and Edit is clicked; a profile is created with the add button in the Health Check list and is selected automatically (how to add). The profile must be of a type that detects the health endpoint returning an error (e.g. HTTP). Details: How should I set the Period, Timeout, Required Successes and Required Failures values?

3

Adım 3 Save the Backend Service (if needed)

The Edit button at the bottom of the form is clicked.

4

Adım 4 Close the health endpoint

The port or page checked on the server is closed so that it returns an error (e.g. the maintenance flag of the application). When the check fails Required Failures times in a row as set in the profile, the server is taken out of traffic.

5

Adım 5 Turn it back on

When the work is finished, the health endpoint is reopened; when the check succeeds Required Successes times in a row, the server returns to traffic.

Verification

1Health statusTR7 side

The Health indicator in the Backend Services list is checked.

ExpectedHealth: Error

Method 6: Maintenance Mode with WEB-CLI

• How it works The maintenance mode in Method 1 is enabled from the command line; with update-many, it can be applied to many Backend Services at once.
• Where it is opened TR7 WEB-CLI → lb backend update and lb backend update-many
• Requirement WEB-CLI access and the ID of the Backend Service
• vService type HTTP and L7 TCP

Step-by-Step Configuration

1

Adım 1 Find the ID

The IDs of the Backend Services are listed:

lb backend list

2

Adım 2 Put a single record into maintenance

The ID seen in step 1 is written in place of <id>:

lb backend update id <id> --maintenanceMode true

3

Adım 3 Put many records into maintenance (if needed)

The bulk update command is written; since no ID is given, a selection list with checkboxes opens, the records are checked, and the warning that states the number of affected records is confirmed by typing the confirmation word:

lb backend update-many target selected --maintenanceMode true

4

Adım 4 End the maintenance

When the work is finished, the same command is run with --maintenanceMode false. In a cluster, the change is transferred to the peer device according to the rule in Method 1.

Verification

1RecordWEB-CLI

The maintenance field in the lb backend show <id> output is checked; the ambulance icon appears in the panel.

ExpectedMaintenance field true

Common Mistakes

• Setting Weight to 1 and assuming the traffic will stop. Weight 1 is the lowest share; it is not zero.
• Setting Connection Limit to 1. The server still receives connections; when the limit is reached, requests wait in the queue.
• Deleting the Backend Service for temporary work. Deletion is permanent; the Backend Service is removed from all the vServices in which it is used and must be set up again together with its settings.
• Turning on the maintenance mode of the group for a single server in the group. All the servers of the group are taken out of traffic; Method 3 is used.
• Trying to verify maintenance mode by looking at the Health column. The Health column does not show the maintenance status; the ambulance icon and Live Stats are checked.
• Relying on Maintenance Mode on a Network vService. Maintenance mode is not applied to the distribution of a Network vService; Method 2 or 3 is used.

How it is done in the documentation

• Backend Services · Maintenance Mode
• Backend Services · Weight
• Backend Groups · Maintenance Mode
• WEB-CLI · lb backend · Bulk Update

Related questions

• What does Maintenance Mode do?
• What is Weight, and how should it be set?
• How do I see which vServices use a Backend Service ("Used in")?