Users

All Posts (397)

avatar

William Alphonso

Database retention policies don't work

avatar
avatar
avatar
avatar

Hi Daniel,

Great catch, and thanks for pulling the exact queries — that pins it down. You're right that the date math is wrong: the scheduled run built the archive window with the upper bound as roughly today plus two years (< 2028-08-29) instead of today minus six months, so it swept in everything rather than only rows older than your six-month setting. That's the same add-instead-of-subtract error, now on the product's automatic path — interestingly, the manual "Clean up now" computes the cut-off correctly, so the two paths differ. I've added this to the report going to our developers alongside the date-format issue. Reverting your settings for now is the right move.

On LogMessage_Old and LogMessage_Old_Archive: yes, you can clear those manually. They're legacy tables that no longer receive entries and aren't part of the retention feature's managed set (that's why they were skipped), so they won't be touched automatically. I'd just take a quick database backup first, then clear them.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Thank you for the confirmation. Our development team are already working on a fix for this issue that will be made available in a 2026.2 revision. I will let you know as soon as we have a specific version for the fix.

Again thank you for your time spent on troubleshooting this!

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hi Daniel,

This is excellent detective work — you've essentially found the root cause, and it's on our side, not your configuration.

You're right about the date format. The cleanup builds its archive/delete statements with a datetime literal like '2025-08-27 00:00:00.000', and because every SQL login on your instance defaults to German (dateformat dmy), SQL Server can't convert that string and terminates the statement with error 242 before any rows are touched. Your SSMS test proves it cleanly: the space-separated form fails under German, while the ISO 8601 form with the T works. That's why nothing was ever archived or deleted, regardless of how large or small the tables were — each table's statement was aborting on the date conversion every time.

To your question about why this never showed up before: those SQL errors are written to the Windows Application Event Log under the DVLS source, which is a different place from the server logs and scheduler traces we'd been looking at. So the nightly automatic runs were most likely failing exactly the same way all along — we just weren't looking at that log. If you want to confirm it, check whether those error-242 entries in the Application log go back before today's "Clean up now" click; I'd expect they do.

For an immediate workaround while we get the product fixed, you can set the default language of the SQL logins DVLS uses to English, which makes that date literal parse correctly. I'd suggest verifying it in your own session first, the same way you tested the failure:

SET LANGUAGE us_english;
SELECT CAST ('2026-07-27 00:00:00.000' AS datetime);

If that returns a valid datetime, then applying it to the logins is the fix:

ALTER LOGIN [your_scheduler_sql_login] WITH DEFAULT_LANGUAGE = us_english;
ALTER LOGIN [your_dvls_app_sql_login]  WITH DEFAULT_LANGUAGE = us_english;

The setting only takes effect on new connections, so restart the Devolutions Scheduler service (and recycle the DVLS app pool) afterward, then try "Clean up now" again. The automatic run should then work at 02:00 as well. Both the scheduled job and "Clean up now" need this, which is why I've included both the scheduler login and the app login.

Separately, you also spotted the "VIEW SERVER STATE denied" line. That one isn't fatal — it only disables a "don't run while the server is busy" safety check, so the cleanup proceeds without it. If you'd like that check to work you can grant VIEW SERVER STATE to the login, but it isn't required for cleanup to function. I'm flagging it to our team along with the date-format issue.

I'm writing this up for our developers now as a product bug, since the cleanup should be using a culture-independent date format rather than depending on the login's language. Thanks again — the SSMS repro you did makes this very easy to hand off.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hi Daniel,

Thanks for sending the scheduler and server logs, and for all the legwork on this one — catching the sign error in the script, building out the full table list, and posting the before/after overviews made this much easier to reason about.

I went through both service logs covering a full day, including the 02:00 window, and here's what stands out. The scheduler's cleanup controller (DatabaseCleanupStartStopTask) is firing on schedule every 30 seconds, but it completes in about a millisecond every time and never actually launches the cleanup worker that does the archiving and deleting — not even at 02:00. There are also no errors anywhere in the logs. So I want to walk back my earlier suggestion that this was the tables being too large to process on the first run: with the worker never starting at all, and no error to show for it, size isn't what's blocking this. The scheduled job simply isn't being triggered. Now that you've trimmed the tables, that also means an automatic run should only ever touch a handful of rows and finish in seconds, which makes the next few checks quick.

Could you try these, in order:

  1. Go ahead and click "Clean up now" once. You were holding off to verify the automatic path, which was the right instinct, but at this point it's our most useful test — with the tables trimmed it'll finish almost instantly, and it tells us whether the in-product cleanup engine works when triggered directly (versus the SQL path you ran manually).

  2. Re-arm the schedule: toggle "Enable automatic cleanup" off, Save, then back on, and Save again. This forces the controller to see a configuration change.

  3. Restart the Devolutions Scheduler service.

  4. Let it sit past the next 02:00, then pull the scheduler log for that window and look for any DatabaseCleanup entry that is not the StartStopTask — that would be the actual worker running. Its presence or absence is the deciding piece.

One more thing that'll help us read the result cleanly: could you confirm from your overview query that there are currently rows older than your archive threshold (one year)? That way, if the automatic run reports nothing moved, we can tell whether it genuinely failed to fire versus there simply being nothing eligible yet.

If "Clean up now" works but the 02:00 run still won't launch the worker after the re-save and service restart, that points to something on the product side rather than your configuration, and I'll take it to our team with your logs attached.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Our development team confirmed that it is safe to delete the records from the UserBehaviorAnalyticsNotification and UserBehaviorAnalyticsNotificationAcknowledger tables.

Because these are a lot more sensitive, we are not including them into the automatic cleanup.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Thank you for the confirmation. I verified and there is no way of deleting them from the Devolutions Server interface. It might be possible to remove them from the DB but I would need to confirm with our development team first. I will let you know once I have a concrete answer.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Thank you for checking this. Would it be possible to configure a log path for the scheduler under? Administration > Server Settings > Logging and also enable the Log debug information (note that this setting should only be enabled temporarily as it can create numerous logs in the DB).

Then wait for the automatic cleanup to run and retrieve the Scheduler logs as well as the Server logs (Data source logs if you are on an older version of DVLS) from the reports tab of Devolutions Server. The Log debug information allows us to log more information for troubleshooting, so the error messages from the Server logs (Data source logs) might change or you might have more information available there.

Once you have those files you can either send them through a ZIP file in direct messages from the forum or email them at service@devolutions.net with FO-55873 in the subject of your email.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Thank you for the confirmation.

Now we just need to keep an eye out to see if the automatic cleanup runs normally. I would suggest capturing the number of rows in the Archived tables and see if the number changes in a couple of days.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Thank you for the confirmation.

For the Archived logs, I would suggest archiving earlier than 2 years. Archived logs are still accessible but they will only be loaded when required instead of being loaded when you open a vault.

When you open a vault, you are loading the entries and all logs attached to these entries that are not archived. By archiving these logs (let's say older than 1 month as an example), we can accelerate the process of opening vaults. Then when you open the log tab on an entry and select a filter that contains logs older than 1 month, then archived logs will be loaded for that specific entry only.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

I sent you a revised script with everything needed to run a test and the full cleanup.

Please let me know if this new version helps.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Sorry about the issue with the script. Yes, I would suggest cleaning any table that has an _Archive version. My script was done some time ago and new tables were added to logs so it is possible that some might be missing from my original script. I will look at cleaning it up.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Thank you for the confirmation.

You do not have to need to configure the scheduler log path unless you are actively troubleshooting the scheduler service.

I think it is highly possible that the scheduler service in failing to run the cleanup because some tables are a bit too big. We would recommend cleaning up the tables manually first.

I will send you a direct message with a SQL query to do a first manual cleanup of the logs, and then the scheduler should be able to run it automatically afterwards.

Best regards,

Database retention policies don't work

avatar
avatar
avatar
avatar

Hello,

Could you please verify in the server logs in DVLS to see if you might have some logs related to the failure of the cleanup?

The scheduler might be unable to delete or archive the logs because there might be too many for the first run.

Would it be possible to run a Disk usage by top table report in SSMS and share it with us through email at service@devolutions.net?

Best regards,

CyberArk Workspace session types

avatar
avatar
avatar

Hello,

Thank you for reaching out!

My name is William and I'm here to assist you in any way I can.

You are correct, the CyberArk Workspace is quite limited compared to the CyberArk Dashboard or CyberArk PVWA entries when using a Devolutions Server.

The CyberArk Workspace is a lot more like using PSM Connect than Remote Desktop Manager (RDM).

Best regards,

Delinea Secret Server

avatar
avatar
avatar
avatar
avatar

Hi David,

Thanks for the clear write-up and screenshots — that comparison between the RDM entry and the Secret Server launcher is exactly what pinned this down. And thank you to Olivier as well; his setup points at the right approach here.

The key detail is in your two screenshots. In RDM, the SSH entry keeps Host and Port as two separate fields (test.something and 3001), whereas Secret Server's own proxied launcher takes the target as a single string — test.something:3001 — with the port riding inside the host value. When RDM hands the session off to the Delinea proxy, the target endpoint (including the port) is resolved on the Delinea side rather than from RDM's Port field, and in this proxied path the separate Port value isn't being carried through — so the proxy falls back to its SSH default of 22. Your own test is the proof that the proxy itself can absolutely reach 3001; it's just a matter of getting that port to the proxy from RDM.

There are two ways to get there. The most robust one is what Olivier described: store the target machine and its port on the Delinea Secret and enable the proxy at the Secret level, then set the credential's resolving mode to "Connection through proxy" in RDM. That way the port lives where the proxied flow expects to resolve it, and RDM doesn't need to pass it at all. If you'd rather not restructure things, it's also worth a quick test to put test.something:3001 directly in RDM's Host field and leave the Port at its default — that mirrors exactly what works in the Secret Server launcher. I'd treat that second one as a "try it and see," since I haven't confirmed how RDM parses a host:port string in that field.

To make sure I'm advising on the right configuration, could you let me know: is this entry a standalone SSH session linked to a Delinea credential, or a Secret-linked entry? And which RDM version are you on? There's been recent work in this area — a new "hybrid" resolving mode shipped in 2026.2.8 — so the version matters.

On the bug-vs-feature question you raised: it sits in a bit of a grey area. In proxy mode the endpoint is meant to be resolved on the Delinea side, so one could argue RDM ignoring its own Port field is consistent. But RDM does show you a Port field that silently does nothing when proxied, and I think that's worth a closer look on our end. Once you confirm the version and entry type, I'll take the "Port field ignored on proxied connections" behaviour back to the team for review.

Talk soon,

Rust Desk Entry Detail

avatar
avatar
avatar
avatar

Bonjour,

Désolé pour le malentendu, le client utilise Devolutions Cloud et non Devolutions Server. J'ai déjà bougé le forum sous Devolutions Cloud mais je ne peux pas enlever le Tag implémenté.

@Marc-Andre Bouchard FYI.

Cordialement,

CyberArk PSM connection workflow clarification

avatar
avatar
avatar

Hello Daniil,

Thank you for the detailed follow-up — this is a good edge case to document.

The cleanest solution, if your CyberArk environments support it, is to set the Authentication on the PVWA credential entry to SAML. With SAML, there are no stored credentials at all — each user authenticates through their identity provider session at launch time. One shared PVWA entry per environment, no per-user duplication required.

If SAML isn't available in all your environments, the workaround is to have each user create their own CyberArk PVWA credential entry in their User Vault — one per CyberArk environment — and then link the shared RDP entries to those User Vault PVWA entries using Find by Name. The naming convention will need to be consistent across users for the lookup to work reliably. It's more maintenance overhead, but it gives each user per-environment credentials without putting them in a shared entry.

As a side note, we don't have plans to add User Vault credential sources directly to the PVWA entry's authentication in the way you're describing. If this is a workflow you'd like to see supported, I'd encourage you to submit a feature request through the forum or via our feedback portal — that's the best way to get it tracked.

Best regards,

Password rotation time different than configured

avatar
avatar
avatar

Hello,

Thank you for the confirmation, I'm glad to know the issue is resolved.

I'll go ahead and mark this thread as resolved.

Best regards,

CyberArk PSM connection workflow clarification

avatar
avatar
avatar

Hello Daniil,

Thank you for reaching out.

My name is William, and I’ll be assisting you with this.

The CyberArk PSM Server and CyberArk PSM Connection entries are no longer the recommended approach for launching sessions through CyberArk PSM.

The recommended approach is to use the CyberArk Dashboard with standard RDP entries.

In this workflow, you would:

  1. Open and authenticate to the CyberArk Dashboard.

  2. Navigate to the desired safe/account.

  3. Select the desired account in the Dashboard.

  4. Launch the standard RDP entry.

Please also ensure that the Connect using dashboard on double-click option is enabled in the RDP entry properties under the Advanced tab. When you double-click the RDP entry, RDM should prompt you to select the connection component to use. You can then select the appropriate PSM component, and the session should launch through CyberArk PSM.

Alternatively, you can use a CyberArk PVWA credential entry instead of the Dashboard. With this entry type, you can set the Resolving mode to PSM Connection and link the credential entry to a standard RDP entry. When launching the RDP entry, RDM will use the account configured in the CyberArk PVWA credential entry and route the session through PSM if the resolving mode is configured accordingly.

For more information, you can refer to the following documentation pages:

Regarding the Connect Using option not being available when you are not authenticated to the CyberArk Dashboard, this is expected. RDM only offers this option when a CyberArk Dashboard is opened and authenticated successfully.

Best regards,

RDM unable to access Hub/Cloud datasource after restart

avatar
avatar
avatar
avatar
avatar

Hello Joe,

Thank you for the detailed report and screenshots.

I was able to reproduce the same behaviour on our side: the Hub/Cloud data source can be added successfully, but after RDM is restarted, the saved data source URL appears to be converted to an unexpected subdomain-based format and then fails with a 404 / “Hub Creating or Not Found” response.

I have reported this to our QA team for investigation. At this point, this looks like an issue with how RDM is saving or reloading the Hub data source URL after restart, rather than an issue with your Hub permissions or account access.

For now, the only workaround I can suggest is to recreate the Hub/Cloud data source when access is required, or use the Hub web interface directly until we have a fix available. I know that is not ideal, but I wanted to be transparent that there does not appear to be a permanent client-side workaround at the moment.

I have attached your Hub/workspace IDs, screenshots, error message, and stack trace to the internal report. We will follow up once we have more information from QA or development.

Best regards,

set default credentials for root object

avatar
avatar
avatar
avatar

Hi G.,

Thank you for testing this and for the details.

You are correct. The error is caused by this part of the script:

[Devolutions.Generated.Models.Connection]::MyDefaultCredentialConnectionID

That type may not be available in the Devolutions Server PowerShell session, depending on the PowerShell context/module being used.

Please use the well-known GUID for My personal credentials instead:

# Set credential configuration to "My personal credentials"
$rootEntry.CredentialConnectionID = "9F3C3BCF-068A-4927-B996-CA52154CAE3B"

So the full example would be:

# Create the vault
$vaultName = "My new vault"
New-DSVault -Name $vaultName

# Retrieve the vault
$vault = Get-DSVault -All | Where-Object { $_.Name -eq $vaultName } | Select-Object -First 1

# Get the root entry of that vault
$root = Get-DSRootSession -VaultID $vault.Id

# Retrieve the root entry as an RDM connection object
$rootEntry = Get-DSEntry -EntryId $root.Id -VaultID $vault.Id -AsRDMConnection

# Set credential configuration to "My personal credentials"
$rootEntry.CredentialConnectionID = "9F3C3BCF-068A-4927-B996-CA52154CAE3B"

# Save the change back to Devolutions Server
Update-DSEntryBase -FromRDMConnection $rootEntry

Sorry for the confusion in the previous example. The original line works in some RDM PowerShell contexts, but for this Devolutions Server workflow, using the well-known GUID directly is the better approach.

Best regards,

Multiple Privileged Accounts?

avatar
avatar
avatar

Hi,

Thank you for the detailed explanation.

For this use case, you should not need to rely on User-Specific Settings for each entry. A cleaner approach would be to configure the shared entries to use Find by name (User vault) as the credential source.

For example, on the shared RDP/SSH entry for CustomerA, you can set the credential mode to: Find by name (User vault)

Then, in the credential name field, enter something like: CustomerA - Admin

Each technician would then create their own credential entry in their own User Vault using that exact same name: CustomerA - Admin

When the shared entry is launched, Remote Desktop Manager will search the current user’s User Vault for a credential matching that name. This means each technician can use their own private admin credential without exposing it to colleagues, while the shared RDP/SSH entries remain common to the team.

You could then standardize this naming convention per customer, for example:

CustomerA - Admin

CustomerB - Admin

CustomerC - Admin

The important part is that the credential entry name expected by the shared entry matches the credential entry name created in each user’s User Vault.

This should avoid the need to store personal admin credentials in the shared customer folder, and should also reduce the amount of per-user configuration required.

Best regards,

Fortinet IPSEC VPN without RDM as admin?

avatar
avatar
avatar
avatar
avatar

Hello Daniel,

Thank you for the details and screenshots.

Based on what you are trying to do, I would not recommend treating the FortiClient IPsec entry as a generic front end for the regular FortiClient VPN interface. The empty Tunnel list means that RDM is not detecting a compatible FortiClient IPsec tunnel it can control through that integration.

The main limitation here does not appear to be that RDM is not running as administrator. The larger issue is that recent FortiClient versions no longer expose the same credential-injection behaviour for IPsec connections. In practice, the seamless username/password injection that works with FortiSSL does not currently translate cleanly to regular FortiClient IPsec.

The supported workaround we normally suggest is to use FortiClient VPN/SSL with the FortiSSL mode and the Fortinet tools package, when SSL VPN is an option. Since you mentioned that SSL VPN through the Fortinet CLI works, that is currently the best-supported path for a seamless RDM workflow.

If IPsec is mandatory, the realistic options are:

  1. Let FortiClient handle the VPN connection directly, including saved credentials or autoconnect if allowed by your FortiGate/FortiClient policy.

  2. Use a Custom VPN (CMD) or Custom VPN (PowerShell) entry only if Fortinet can provide a supported command line for your exact FortiClient Windows version.

  3. Launch FortiClient from RDM, but complete the IPsec authentication in FortiClient manually.

At this point, I would not expect RDM to inject the username/password/key into the regular FortiClient IPsec UI in the same way it can with FortiSSL.

Could you confirm the FortiClient version installed on the workstation and whether the VPN profile is locally configured or provisioned through FortiClient EMS? That will help us confirm whether there is any supported command-line option available for your specific setup.

Best regards,

set default credentials for root object

avatar
avatar
avatar
avatar

Hello G.,

Thank you for your question.

I can confirm that there are currently no Devolutions Server setting to automatically set My personal credentials as the default credential configuration for every newly created vault.

However, this can be done with PowerShell after the vault is created by updating the root entry of the vault. Here is an example:

# Create the vault
$vaultName = "My new vault"
New-DSVault -Name $vaultName

# Retrieve the vault
$vault = Get-DSVault -All | Where-Object { $_.Name -eq $vaultName } | Select-Object -First 1

# Get the root entry of that vault
$root = Get-DSRootSession -VaultID $vault.Id

# Retrieve the root entry as an RDM connection object
$rootEntry = Get-DSEntry -EntryId $root.Id -VaultID $vault.Id -AsRDMConnection

# Set credential configuration to "My personal credentials"
$rootEntry.CredentialConnectionID = [Devolutions.Generated.Models.Connection]::MyDefaultCredentialConnectionID

# Save the change back to Devolutions Server
Update-DSEntryBase -FromRDMConnection $rootEntry

Please make sure the PowerShell session is already authenticated to your Devolutions Server instance and that the account running the script has sufficient permissions to create and modify vaults.

Best regards,

Devolutions Server & Gateway side-by-side via Azure Application Proxy

avatar
avatar
avatar

Hello Dave,

Thank you for taking the time to document the setup and behaviour here.

Since we already have a session scheduled together, I am assigning myself this thread as well so I can keep both discussions aligned. We will follow up here after our session with any findings or recommendations from the review.

In the meantime, I will leave the thread open in case anyone else from the community has additional input or has tested a similar setup with Azure Application Proxy and a side-by-side Devolutions Gateway deployment.

Best regards,

1 - 25 of 397 items