Troubleshooting
This section contains information about how to monitor the system and how to troubleshoot specific issues that may occur.
This section contains information about where information can be found within RES.
There are various sources of information retained that can be referenced for troubleshooting and monitoring uses.
Log files exist on the EC2 instances in use by RES. The SSM Session Manager can be used to open a session to the instance for examining these files.
On infrastructure instances such as the cluster-manager and vdc-controller, application and other logs can be found at the following locations.
- /opt/idea/app/logs/application.log
- /root/bootstrap/logs/
- /var/log/
- /var/log/sssd/
- /var/log/messages
- /var/log/user-data.log
- /var/log/cloud-init.log
- /var/log/cloud-init-output.log
On a Linux virtual desktop, the following contain useful log files
- /var/log/dcv/
- /root/bootstrap/logs/userdata.log
- /var/log/messages
On Windows virtual desktop instances logs can be found at
- PS C:\ProgramData\nice\dcv\log
- PS C:\ProgramData\nice\DCVSessionManagerAgent\log
On Windows, some applications logging can be found at:
- PS C:\Program Files\NICE\DCV\Server\bin
On Windows, the NICE DCV certificate files can be found in:
- C:\Windows\System32\config\systemprofile\AppData\Local\NICE\dcv\
CloudWatch Log Groups
The EC2 and Lambda compute resources will log information to CloudWatch Log Groups. The log entries within them can provide useful information when troubleshooting potential issues or for general information.
Those groups will be named as follows:
- /aws/lambda/< envname >-/ - lambda related
- /< envname >/
- analytics/ - OpenSearch database related
- cluster-manager/ - main infrastructure host
- vdc/ - virtual desktop related
- dcv-broker/ - desktop related
- dcv-connection-gateway/ - desktop related
- controller/ - main desktop controller host
- dcv-session/ - desktop session related
When examining log groups, it can be helpful to filter using upper and lower case strings such as the following. This will output only those messages containing the noted strings.
?"ERROR" ?"error"
Another method of monitoring for issues is to create CloudWatch Dashboards that contain widgets displaying the data of interest.
An example is to create a widget that counts the occurrence of the strings error and ERROR and graph them as lines. This method makes it easier to detect the occurrence of potential issues or trends indicating a pattern change has occurred.
The following is an example of that for the infrastructure hosts. To use this, concatenate the query lines and replace the < envname > and < region > attributes with the appropriate values.
{
"widgets": [
{
"type": "log",
"x": 0,
"y": 0,
"width": 24,
"height": 6,
"properties": {
"query": "SOURCE '/<envname>/vdc/controller' |
SOURCE '/<envname>/cluster-manager' |
SOURCE '/<envname>/vdc/dcv-broker' |
SOURCE '/<envname>/vdc/dcv-connection-gateway' |
fields @timestamp, @message, @logStream, @log\n|
filter @message like /(?i)(error|ERROR)/\n|
sort @timestamp desc|
stats count() by bin(30s)",
"region": "<region>",
"title": "infrastructure hosts",
"view": "timeSeries",
"stacked": false
}
}
]
}
An example of the Dashboard might appear as follows.
The CloudFormation stacks created during environment creation contain resources, event, and output information associated with the configuration of the environment. For each of the stacks, the Events, Resources, and Outputs tab can be referred to for information about the stacks.
Batteries Included (BI) stacks
- < envname >
- < envname >-Networking
- < envname >-DirectoryService
- < envname >-Storage
- < envname >-WindowsManagementHost
RES stacks
- < envname >-bootstrap
- < envname >-cluster
- < envname >-metrics
- < envname >-directoryservice
- < envname >-identity-provider
- < envname >-analytics
- < envname >-shared-storage
- < envname >-cluster-manager
- < envname >-vdc
- < envname >-bastion-host
If the RES UIs are indicating server errors, the cause may be an application software or other issue.
Each of the infrastructure EC2 instance autoscaling groups (ASGs) contain an Activity tab that can be useful for detecting scaling activity for the instances. If UI pages note any errors or are not accessible, check the EC2 console for multiple terminated instances and check the Auto Scaling Group Activity tab for the related ASG to determine if EC2 instances are cycling.
If so, use the related CloudWatch log group for the instance to determine if errors are being logged that might indicate the cause of the issue. It may also be possible to use the SSM Session console to open a session to a running instance of that type and examine the log files on the instance to determine a cause before the instance is marked as unhealthy and terminated by the ASG.
The ASG console may show activity similar to the following if this issue is occurring.
This section contains screenshots of the system operating in various states.
The EC2 console, when no desktops are running, will typically look similar to the following. The instances that are show are the RES infrastructure EC2 hosts. The prefix in the name of the instances will be the RES environment name.
The EC2 console when virtual desktops are running, they will appear similar to the following. In this case, the virtual desktops are noted in red. The suffix to the instance name will be the user that created the desktop. The name in the center will be the Session Name set at launch time and will be either be the default MyDesktop or the name set by the user.
When the EC2 console shows terminated instances, they will generally be desktop hosts that have been terminated. If the console includes infrastructure hosts in a terminated state, particularly if there are multiple of the same type, that may indicate a system issue in progress.
The following show desktop instances that have been terminated.
The following are examples of ldap related commands that can be entered on infrastructure hosts to view AD configuration related information. The domain and other parameters used should reflect those entered at environment creation time.
ldapsearch "(cn=AWS Delegated Add Workstations To Domain Users)" -x -h corp.res.com
-b "DC=corp,DC=res,DC=com" -D "CN=Admin,OU=Users,OU=CORP,DC=corp,DC=res,DC=com"
-w <password>
ldapsearch "(&(objectClass=group))" -x -h corp.res.com
-b "DC=corp,DC=res,DC=com" -D "CN=Admin,OU=Users,OU=CORP,DC=corp,DC=res,DC=com"
-w <password>
On a Windows desktop, list the session associated with it using the following
PS C:\Windows\System32\config\systemprofile\AppData\Local\NICE\dcv> & 'C:\Program Files\NICE\DCV\Server\bin\dcv.exe'list-sessions
Session: 'a7953489-9dbf-492b-8135-7709dccc4cab' (owner:admin2 type:console name:windows1)
NICE DCV is utilized for virtual desktop sessions. https://aws.amazon.com/hpc/dcv/. The following are examples of determining the version of the DCV software installed.
Linux
[root@ip-10-3-157-194 ~]# /usr/bin/dcv version
NICE DCV 2023.0 (r14852)
Copyright (C) 2010-2023 NICE s.r.l.
All rights reserved.
This product is protected by copyright and
licenses restricting use, copying, distribution, and decompilation.
Windows
PS C:\Windows\System32\config\systemprofile\AppData\Local\NICE\dcv> & 'C:\Program Files\NICE\DCV\Server\bin\dcv.exe' version
NICE DCV 2023.0 (r15065)
Copyright (C) 2010-2023 NICE s.r.l.
All rights reserved.
This product is protected by copyright and
licenses restricting use, copying, distribution, and decompilation.
The following section contains issues that may occur, how to detect them, and suggestions on how to resolve the issue.
If you receive an error that you're not authorized to perform the iam:PassRole action, your policies must be updated to allow you to pass a role to RES.
Some AWS services allow you to pass an existing role to that service instead of creating a new service role or service-linked role. To do this, you must have permissions to pass the role to the service.
The following example error occurs when an IAM user named marymajor tries to use the console to perform an action in RES. However, the action requires the service to have permissions that are granted by a service role. Mary does not have permissions to pass the role to the service.
User: arn:aws:iam::123456789012:user/marymajor is not authorized to perform: iam:PassRole
In this case, Mary's policies must be updated to allow her to perform the iam:PassRole action. If you need help, contact your AWS administrator. Your administrator is the person who provided you with your sign-in credentials.
I want to allow people outside of my AWS account to access my Research and Engineering Studio on AWS resources
You can create a role that users in other accounts or people outside of your organization can use to access your resources. You can specify who is trusted to assume the role. For services that support resource-based policies or access control lists (ACLs), you can use those policies to grant people access to your resources.
To learn more, consult the following:
- To learn how to provide access to your resources across AWS accounts that you own, see Providing access to an IAM user in another AWS account that you own in the IAM User Guide.
- To learn how to provide access to your resources to third-party AWS accounts, see Providing access to AWS accounts owned by third parties in the IAM User Guide.
- To learn how to provide access through identity federation, see Providing access to externally authenticated users (identity federation) in the IAM User Guide.
- To learn the difference between using roles and resource-based policies for cross-account access, see How IAM roles differ from resource-based policies in the IAM User Guide.
The file systems need to be in “Available” state before they can be mounted by VDI hosts. Follow steps below to validate file system is in needed state
EFS
- Go to EFS console [us-east-1]
- Check File system state is Available
- If File system state is not Available, wait before launching VDI hosts
FSx ONTAP
- Go to FSx console [us-east-1]
- Check Status is Available
- If Status is not Available, wait before launching VDI hosts
The file systems onboarded on RES should have the needed security group rules configured to allow VDI hosts to mount the file systems. As these file systems are created externally to RES, RES doesn’t manage the associated security group rules.
The security group associated with the onboarded file systems should allow following inbound traffic
- NFS traffic (port: 2049) from the linux VDC hosts
- SMB traffic (port: 445) from the windows VDC hosts
ONTAP supports UNIX, NTFS and MIXED security style for the volumes. The security styles determine the type of permissions ONTAP uses to control data access and what client type can modify these permissions.
For example, if a volume uses UNIX security style, SMB clients can still access data (provided that they properly authenticate and authorize) due to the multi-protocol nature of ONTAP. However, ONTAP uses UNIX permissions that only UNIX clients can modify using native tools.
Using UNIX style volume with Linux workloads
Permissions can be configured by the sudoer for other users. For eg. following would give all members of <group-ID> full read/write permissions on /<project-name> directory
sudo chown root:<group-ID> /<project-name>
sudo chmod 770 /<project-name>
Using NTFS style volume with Linux and Windows workloads
Permissions can be configured on the file share using share properties. For eg. following would give pentest_user_01 full control permissions on the file share myshare.
If the volume is going to be used by both Linux and Windows client we need to set up a name mapping on SVM that will associate any Linux user name to the same user name with the NetBIOS domain name format of domain\username. This is needed to translate between Linux and Windows users. For reference see Enabling multiprotocol workloads with Amazon FSx for NetApp ONTAP
Currently, if you are creating FSx for NetApp ONTAP from RES console, the file system gets provisioned but it does not join the domain. To join the created ONTAP file system SVM to your domain, please see here Joining SVMs to a Microsoft Active Directory and follow the steps on Amazon FSx console [us-east-1]. Once the SVM joins the domain successfully, go to SVM Summary > Endpoints > SMB DNS name, copy the DNS name as it will be needed later.
After joining to domain, edit the SMB DNS config key in cluster settings DynamoDB table:
- Go to Amazon DynamoDB console [us-east-1] 2.** Tables >** select -cluster-settings
-
Explore table items > expand Filters, and put following filter
- Attribute name - key
- Condition - Equal to
- Value - shared-storage..fsx_netapp_ontap.svm.smb_dns
- Select the returned item > Actions > Edit item
- Update the value to string value (SMB DNS name copied earlier) 6.** Save and close**
Additionally, ensure security group associated with the file system allows traffic as recommended on File System Access Control with Amazon VPC. New VDI hosts using the file system will now be able to mount the domain joined SVM and file system.
Alternative, you may onboard existing file system which is already joined to your domain using RES Onboard File System capability from Environment Management > File Systems > Onboard File System
This situation indicates a misconfiguration of your SSO integration. To determine the issue, check the logs for the controller instance. Do this by:
- See logs
- Opening Log groups section of the AWS CloudWatch Console
- Search for “cluster-manager” and select the one that corresponds to your environment.
- Look for any errors that show up in the logs
- See SSO configuration settings, to know if there was a mistake
- AWS Console → DynamoDB
- Cluster settings table for the corresponding env
- Explore table data
- Filters → key = key, condition = contains, value = sso
- Run
- Verify that the sso configuration values are correct
- If wrong, manually set the sso_enabled key’s value to false in DDB
- Go back to the UI and you will be able to re-configure SSO
Access the env with the parameter sso=false as <env-url>?sso=false
The AD sync needs to be triggered. The AD Sync process is a scheduled event where once an hour, the default schedule, a Lambda runs that performs the sync of the AD with the RES environment.
If changes such as users or groups, there may be a lag until the next AD Sync process executes. To trigger it manually, perform the following at the AWS Lambda console
- search for the Lambda named
< env-name >-cluster-manager-scheduled-ad-sync - select the Test tab
- search for the test event named CloudWatch
- the content should be similar to the following=
{
"id": "cdc73f9d-aea9-11e3-9d5a-835b769c0d9c",
"detail-type": "Scheduled Event",
"source": "aws.events",
"account": "123456789012",
"time": "1970-01-01T00:00:00Z",
"region": "us-west-2",
"resources": [
"arn:aws:events:us-east-1:123456789012:rule/ExampleRule"
],
"detail": {}
}
- save that event to a new name
- select Test to trigger the execution
- the CloudWatch Log group called < env-name >-cluster-manager should show the execution of the sync process
As an admin, I don’t see a user that is a part of Active Directory in the list of users when creating session
Sessions can only be created for active users. Active users are users who have logged into the environment at least once.
2023-10-31T18:03:12.942-07:00 ldap.SIZELIMIT_EXCEEDED: {'msgtype': 100, 'msgid': 11, 'result': 4, 'desc': 'Size limit exceeded', 'ctrls': []}
If this error is appearing in the cluster-manager CloudWatch log it may indicate that too many user records are being returned in an ldap search. The mitigation is to change the configure of the IDP to increase the limit for the number of results returned in an ldap search.
If an invitation email was not received after the CloudFormation stacks were successfully created the cause can be narrowed down by checking the following
Confirm the email address parameter was entered correctly
If the email address is incorrect or the email account cannot be accessed, the RES environment should be deleted and recreated.
Check the EC2 console for evidence of cycling instances
If there are EC2 instances that have the < envname > name prefix are appearing as terminated and being replaced with new instances, there may be an issue with the network or AD configuration.
When using Batteries Included If the Batteries Included stack was used, confirm that the VPC, private, and public subnets, and other parameters selected during the RES installation were created by the BI stack. If other existing resources were used, there may be a issue with the RES resources not being able to communicate with each other, the AD, or other AWS services.
If any of the selections were incorrect, it may be necessary to delete the RES stack and recreate it.
When using existing resources If external existing resources were entered during the RES environment creation, confirm that the networking and AD match the expected configurations.
With either environment creation type, confirming that the infrastructure instances successfully joined the AD domain is critical. An example of a method to do that is to open an SSM Session to the instance and try the following. If no output appears, that indicates an issue with the AD configuration.
corp.res.com type: kerberos realm-name: CORP.RES.COM domain-name: corp.res.com configured: kerberos-member server-software: active-directory client-software: sssd required-package: oddjob required-package: oddjob-mkhomedir required-package: sssd required-package: adcli required-package: samba-common-tools login-formats: %U login-policy: allow-realm-logins
If an environment deletion fails due a dependent object error such as a the vdcdcvhostsecuritygroup, this may be due to an EC2 instance having been launched using the AWS console into a RES created subnet or using a RES created security group. The resolution is to check all EC2 instances to determine if any where launched in this manner, terminate any found, and resume the environment deletion.
When trying to create an environment, this error appeared for the CIDR block provided as a parameter
Received response status [FAILED] from custom resource. Message returned: failed to update cluster prefix list: An error occurred (InvalidParameterValue) when calling the ModifyManagedPrefixList operation: The specified CIDR (52.94.133.132/24) is not valid. For example, specify a CIDR in the following form: 10.0.0.0/16.
The expected format should be x.x.x.0/24 or x.x.x.0/32
How to check the expiration date of a certificate. This example is from an aml2 virtual desktop instance.
root@ip-17-0-128-66 dcv]# openssl x509 -enddate -noout -in /var/lib/dcv/.config/NICE/dcv/dcv.pem
notAfter=Aug 22 17:44:37 2024 GMT
Environment creation involves a number of resources being created through a series of operations. It is possible that, for the region selected, a capacity issue may occur that might cause a CloudFormation stack creation to fail.
If that occurs, potential resolutions can be to delete the environment and retry the create. The issue may be transient and attempting again might successfully create. An alternative is to create the environment in a different region.
If the Batteries Included stack creation fails with an error similar to the following, it may be due to EC2 patching occurring unexpectedly during the provisioning that occurs after instance launch.
AdDomainAdminNode CREATE_FAILED Failed to receive 1 resource signal(s) within the specified duration
Methods to determine the cause of the failure include the following
- check the SSM State Manager to see if patching is configured and if it is set for "all instances"
- check the SSM RunCommand/Automation execution history to see if an execution of a potentially patching related SSM document coincides with an instance launch
- determine from local instance logging if the instance rebooted during provisioning, the instance log files to check are noted in the section Log files on the environment EC2 instances
If patching is determined to be the cause, the remediation is to delay patching for the RES instances for at least 15-minutes after launch.
On the RES Snapshots page, if a snapshot has a status of Failed, the cause can be determined by going to the CloudWatch log group for the cluster-manager for the time that the error occurred.
[2023-11-19 03:39:20,208] [INFO] [snapshots-service] creating snapshot in S3 Bucket:
asdf at path s31
[2023-11-19 03:39:20,381] [ERROR] [snapshots-service] An error occurred while
creating the snapshot: An error occurred (TableNotFoundException)
when calling the UpdateContinuousBackups operation:
Table not found: res-demo.accounts.sequence-config
If issues such as server error messages are appearing in the UI or desktop sessions cannot connect, that may indicate an issue in the infrastructure EC2 instances.
The methods to determine the source of the issue are to first check the EC2 console for any EC2 instances that appear to be repeatedly terminating and being replaced by new instances. If that is the case, checking the CloudWatch logs may determine the cause.
Another method is check the load balancers in the system. An indication that there may be system issues is if any load balancers, found on the EC2 console, do not show any healthy instances registered.
An example of a normal appearance is in the following:
If the Healthy entry is 0, that indicates that no EC2 instance is available to process requests.
If the Unhealthy entry is non-0, that indicates that an EC2 instance may be cycling. This can be due to the installed applications software not passing health checks.
If both Healthy and Unhealthy entries are 0, that indicates a potential network misconfiguration. An example might be the public and private subnets do not have corresponding AZs. If this condition occurs, there may be additional text on the console indicating that network state exists.
If a desktop connection closes or can no longer be connected to the issue may be due to the underlying EC2 instance failing or the EC2 instance has been terminated or stopped outside of the RES environment. The Admin UI status may continue to show a ready state but attempts to connect to it fail.
The AWS EC2 Console should be used to determine if the instance has been terminated or stopped. If stopped, try starting it again. If the state is terminate, another desktop will have to be created. Any data that was stored on the user home directory should still be available when the new instance starts.
If the previous failed instance still appears on the Admin UI, it may need to be terminated using the Admin UI.
The default limit for the number of virtual desktops that a user can launch is 5. This can be changed by an admin using the Admin UI as follows
- go to eVDI Settings
- select the **Server **tab
- in the DCV Session panel, click on the edit icon on the right
- change the value in Allowed Sessions Per User to the desired new value
- click Submit
- refresh the page to confirm that the new setting is in place
If a Windows desktop connection fails with the UI error “The connection has been closed. Transport error”, the cause can be due to an issue in the DCV server software related to certificate creation on the Windows instance.
The CloudWatch log group /vdc/dcv-connection-gateway may log the connection attempt error with messages similar to the following
Nov 24 20:24:27.631 DEBUG HTTP:Splicer Connection{id=9}:
Websocket{session_id="1291e75f-7816-48d9-bbb2-7371b3b911cd"}:
Resolver lookup{client_ip=Some(52.94.36.19)
session_id="1291e75f-7816-48d9-bbb2-7371b3b911cd"
protocol_type=WebSocket extension_data=None}:NoStrictCertVerification:
Additional stack certificate (0): [s/n: 0E9E9C4DE7194B37687DC94D2C0F5E94AF0DD57E]
Nov 24 20:25:15.384 INFO HTTP:Splicer Connection{id=21}:Websocket{
session_id="d1d35954-f29d-4b3f-8c23-6a53303ebc3f"}:
Connection initiated error: unreachable, server io error Custom {
kind: InvalidData, error:
General("Invalid certificate: certificate has expired (code: 10)") }
Nov 24 20:25:15.384 WARN HTTP:Splicer Connection{id=21}:
Websocket{session_id="d1d35954-f29d-4b3f-8c23-6a53303ebc3f"}:
Error in websocket connection: Server unreachable: Server error: IO error:
unexpected error: Invalid certificate: certificate has expired (code: 10)
If this occurs, a resolution may be to use the SSM Session Manager to open a connection to the Windows instance and remove the following 2 certificate related files.
PS C:\Windows\system32\config\systemprofile\AppData\Local\NICE\dcv> dir
`Directory: C:\Windows\system32\config\systemprofile\AppData\Local\NICE\dcv`
Mode LastWriteTime Length Name
---- ------------- ------ ----
-a---- 8/4/2022 12:59 PM 1704 dcv.key
-a---- 8/4/2022 12:59 PM 1265 dcv.pem
The files should be automatically recreated and a subsequent connection attempt may be successful.
If this method resolves the issue and if new launches of Windows desktops produce the same error, use the Create Software Stack function to create a new Windows software stack of the fixed instance with the regenerated certificate files. That may produce a Windows software stack that can be used for successful launches and connections.
If a desktop launch remains i the provisioning state in the Admin UI, this may be due to several reasons.
Methods to determine the cause including examining log files on the desktop instance for errors that might be causing the issue. This document contains a list of log files and CloudWatch log groups that contain relevant information in the section labelled Useful log and event information sources.
The following are potential causes of this issue.
- AMI id used has been registered as a software-stack but is not supported by RES
The bootstrap provisioning script failed to complete because the AMI does not have the expected configuration or tooling required. The log files on the instance, such as /root/bootstrap/logs/ on a Linux instance may contain useful information regarding this.
AMIs ids taken from the AWS Marketplace may not work for RES desktop instances. They require testing to confirm if they are supported.
If an infrastructure instance is repeatedly showing as terminated in the EC2 console, the cause is potentially configuration related. The potential cause may be dependent on the infrastructure instance type. The following are methods to determine the cause.
If the vdc-controller instance is showing repeated terminated states in the EC2 console, this can be due to an incorrect Secret tag. Secrets that are maintained by RES have tags that are used as a part of the IAM access control policies attached to the infrastructure EC2 instances. If the vdc-controller is cycling and the following error appears in the CloudWatch log group, the cause may be that a secret has not been tagged correctly. Note that the secret needs to be tagged with the following:
{
"res:EnvironmentName": "< envname >" # e.g. "res-demo"
"res:ModuleName": "virtual-desktop-controller"
}
The CloudWatch log message for this error will appear similar to the following.
An error occurred (AccessDeniedException) when calling the GetSecretValue
operation: User: arn:aws:sts::160215750999:assumed-role/<envname>-vdc-gateway-role-us-east-1/i-043f76a2677f373d0
is not authorized to perform: secretsmanager:GetSecretValue on resource:
arn:aws:secretsmanager:us-east-1:160215750999:secret:Certificate-res-bi-Certs-5W9SPUXF08IB-F1sNRv
because no identity-based policy allows the secretsmanager:GetSecretValue action
Check the tags on the EC2 instance and confirm that the match the above list.