You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Global setting "consoleproxy.session.timeout " is not honoured #12810
Hi @kiranchavala,
I reviewed the issue and it seems that the global setting consoleproxy.session.timeout is not being applied during console session handling.
I’d like to investigate how this configuration is read and whether it is properly passed and enforced in the console proxy session lifecycle. Based on that, I’ll try to identify where the timeout should be applied.
Please let me know if this approach looks good or if there are specific areas I should focus on.
And can I start working on this or not ?
I have not started creating any experimental code yet. Earlier I asked if anyone was already working on this, but since I did not get a response, I held off on starting anything to avoid duplicating work.
If no one is working on it, I’d like to proceed. My plan is to:
Trace how the consoleproxy.session.timeout global setting is read and propagated to the console proxy (CPVM), and
Identify where in the console session lifecycle the timeout should be enforced and why it is not currently honoured.
If you have any specific components or files you’d recommend focusing on first, I’d be happy to take that into account before starting on a draft fix.
I think this will go in the systemvm code and needs to be propagated from the service in the services/console-proxy project. I would have to search as well.
I think this will go in the systemvm code and needs to be propagated from the service in the services/console-proxy project. I would have to search as well.
Thanks, that clarifies it.
I’ll start working on this issue. My next steps will be:
Look into the SystemVM/CPVM side and the services/console-proxy project to see how console proxy configuration is currently propagated, and
Add/enforce the consoleproxy.session.timeout value in the console proxy session handling so that console sessions respect the configured timeout.
Once I have something concrete, I’ll open a draft PR .
I’ve implemented code changes for this issue so that consoleproxy.session.timeout (in milliseconds) is read in ConsoleProxy, passed into the console proxy process, and then used by both the noVNC WebSocket session handling and the GC thread for idle session cleanup.
I’ve implemented code changes for this issue so that consoleproxy.session.timeout (in milliseconds) is read in ConsoleProxy, passed into the console proxy process, and then used by both the noVNC WebSocket session handling and the GC thread for idle session cleanup.
I’ve implemented code changes for this issue so that consoleproxy.session.timeout (in milliseconds) is read in ConsoleProxy, passed into the console proxy process, and then used by both the noVNC WebSocket session handling and the GC thread for idle session cleanup.
The changes are in the following files:
services/console-proxy/server/src/main/java/com/cloud/consoleproxy/ConsoleProxy.java
services/console-proxy/server/src/main/java/com/cloud/consoleproxy/ConsoleProxyNoVNCHandler.java
services/console-proxy/server/src/main/java/com/cloud/consoleproxy/ConsoleProxyGCThread.java
Is it possible to get this fix back ported to the 4.22 LTS version? It is my understanding 4.23 is not LTS.
Regarding backporting to 4.22 LTS: I’m not a release manager, so I can’t decide that myself, but once the PR is reviewed and merged the maintainers can consider backporting it to the 4.22 branch if it meets their criteria.
Activity
dheeraj12347 commentedon Mar 18, 2026
Hi @kiranchavala,
I reviewed the issue and it seems that the global setting
consoleproxy.session.timeoutis not being applied during console session handling.I’d like to investigate how this configuration is read and whether it is properly passed and enforced in the console proxy session lifecycle. Based on that, I’ll try to identify where the timeout should be applied.
Please let me know if this approach looks good or if there are specific areas I should focus on.
And can I start working on this or not ?
DaanHoogland commentedon Apr 1, 2026
@dheeraj12347 , are you creating (experimental) code for this?
dheeraj12347 commentedon Apr 1, 2026
@DaanHoogland Thanks for the follow-up.
I have not started creating any experimental code yet. Earlier I asked if anyone was already working on this, but since I did not get a response, I held off on starting anything to avoid duplicating work.
If no one is working on it, I’d like to proceed. My plan is to:
consoleproxy.session.timeoutglobal setting is read and propagated to the console proxy (CPVM), andIf you have any specific components or files you’d recommend focusing on first, I’d be happy to take that into account before starting on a draft fix.
DaanHoogland commentedon Apr 1, 2026
I think this will go in the systemvm code and needs to be propagated from the service in the
services/console-proxyproject. I would have to search as well.dheeraj12347 commentedon Apr 2, 2026
Thanks, that clarifies it.
I’ll start working on this issue. My next steps will be:
services/console-proxyproject to see how console proxy configuration is currently propagated, andconsoleproxy.session.timeoutvalue in the console proxy session handling so that console sessions respect the configured timeout.Once I have something concrete, I’ll open a draft PR .
vishesh92 commentedon Apr 8, 2026
@dheeraj12347 Let me know if you need any help with fixing this issue.
dheeraj12347 commentedon Apr 8, 2026
I’ve implemented code changes for this issue so that consoleproxy.session.timeout (in milliseconds) is read in ConsoleProxy, passed into the console proxy process, and then used by both the noVNC WebSocket session handling and the GC thread for idle session cleanup.
The changes are in the following files:
services/console-proxy/server/src/main/java/com/cloud/consoleproxy/ConsoleProxy.java
services/console-proxy/server/src/main/java/com/cloud/consoleproxy/ConsoleProxyNoVNCHandler.java
services/console-proxy/server/src/main/java/com/cloud/consoleproxy/ConsoleProxyGCThread.java
vishesh92 commentedon Apr 9, 2026
That's good. Is there a draft PR, I can check?
dheeraj12347 commentedon Apr 11, 2026
@vishesh92 I have drafted the PR :)
DRossII commentedon Apr 17, 2026
Is it possible to get this fix back ported to the 4.22 LTS version? It is my understanding 4.23 is not LTS.
dheeraj12347 commentedon Apr 18, 2026
Regarding backporting to 4.22 LTS: I’m not a release manager, so I can’t decide that myself, but once the PR is reviewed and merged the maintainers can consider backporting it to the 4.22 branch if it meets their criteria.
dheeraj12347 commentedon Apr 18, 2026
If the fix is accepted, I’m happy to help by opening a backport PR against the 4.22 branch if maintainers think that’s appropriate.
DaanHoogland commentedon Apr 21, 2026
I think it is appropriate @dheeraj12347 go ahead and rebase it. I would actually suggest the 4.20 branch
dheeraj12347 commentedon Apr 22, 2026
Thanks, @dahn.
I’ve rebased this onto the
4.20branch and opened the backport PR as suggested.This PR now targets
apache:4.20(backport-12810-consoleproxy-timeout-4.20on my fork) and builds cleanly:Please let me know if you’d like any further adjustments.