Memoryleak

Hi, @piyush.rajput , Thanks for details. We were able to create an updated self-hosted image where it frees system memory after each conversion, we need some time to test and publish it and then will update you soon.

1 Like

Hi, @piyush.rajput, We have published an update. Please update GroupDocs.Conversion Cloud image to ‘latest’ or ‘26.4.4’ tag. And please be sure to set new parameter: ISOLATED_CONVERSION=true.

1 Like

Hi @sergei.terentev ,
Pods still restarting with OOM, here is the pod last state

    State:          Running
      Started:      Mon, 20 Jul 2026 22:43:01 +0530
    Last State:     Terminated
      Reason:       OOMKilled
      Exit Code:    137
      Started:      Mon, 20 Jul 2026 18:14:07 +0530
      Finished:     Mon, 20 Jul 2026 22:43:00 +0530
    Ready:          True
    Restart Count:  1
    Limits:
      cpu:     8
      memory:  12Gi
    Requests:
      cpu:     1
      memory:  1Gi
    Liveness:  http-get http://:80/ delay=10s timeout=5s period=30s #success=1 #failure=3
    Environment:
      LICENSE_PUBLIC_KEY:   <set to the key 'LICENSE_PUBLIC_KEY' in secret 'groupdocs-secrets'>   Optional: false
      LICENSE_PRIVATE_KEY:  <set to the key 'LICENSE_PRIVATE_KEY' in secret 'groupdocs-secrets'>  Optional: false
      ISOLATED_CONVERSION:  true

Hi, @piyush.rajput,
For better understanding the issue cause, can you please

  1. Update the image to latest or 26.4.5 tag, keep ISOLATED_CONVERSION enabled
  2. Collect and share extended logs which was added to this version

Here is some details about collecting logs:

Collecting logs after an OOMKill

When a pod is OOMKilled, Kubernetes restarts it immediately and the container’s stdout is reset — logs from before the crash are no longer available via the default kubectl logs command. Use --previous to read the terminated container’s last log buffer:


# Replace <pod-name> and <namespace> with your values

kubectl logs --previous <pod-name> -n <namespace>

To find the pod name:


kubectl get pods -n <namespace>

What to look for

The service logs a summary line after every completed conversion:


info: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]

Conversion #42 | docx -> pdf |input 8503 KB | avg input 5222 KB | WorkingSet 9821 MB (80% of 12288 MB) | GC heap 312 MB 

Share the last 50–100 lines from --previous output so we can see which conversion number caused the spike and how memory was trending.

HI @sergei.terentev ,
We are not getting logs from the groupdocs image. Getting only following previous/current logs

info: GroupDocs.Conversion.Cloud.Web.Startup[0]
      The license has been set.
Hosting environment: Production
Content root path: /app
Now listening on: http://[::]:80
Application started. Press Ctrl+C to shut down.

Hi, @piyush.rajput,
I have updated the image and added a log, indicating isolated mode is enabled (and logs also). So can you please update image and try again. Thanks.

1 Like

Hi @sergei.terentev ,
There are still no logs from the pod. These are the only logs:

2026-07-22 00:27:31.581       The license has been set.
2026-07-22 00:27:31.738 Hosting environment: Production
2026-07-22 00:27:31.738 Content root path: /app
2026-07-22 00:27:31.738 Now listening on: http://[::]:80
2026-07-22 00:27:31.738 Application started. Press Ctrl+C to shut down.

Hi, @piyush.rajput,
Please ensure you have correctly updated image, the log should have isolated conversion status:

info: GroupDocs.Conversion.Cloud.Web.Startup[0]
      The license has been set.
info: GroupDocs.Conversion.Cloud.Web.Startup[0]
      Isolated conversion mode enabled
...

If isolated conversion disabled, it should write: “Isolated conversion mode disabled”
Missing message probably means you have not updated the image from the dockerhub

Hi @sergei.terentev ,
I have re-built/fetched the latest image and then deployed at 10:27PM IST. Still seeing same logs. Here is the state details from k8s (memory has spiked, check the screenshot):

 State:          Running
      Started:      Tue, 21 Jul 2026 22:27:51 +0530
    Ready:          True
    Restart Count:  0
    Limits:
      cpu:     8
      memory:  12Gi
    Requests:
      cpu:     1
      memory:  1Gi
    Liveness:  http-get http://:80/ delay=10s timeout=5s period=30s #success=1 #failure=3
    Environment:
      LICENSE_PUBLIC_KEY:   <set to the key 'LICENSE_PUBLIC_KEY' in secret 'groupdocs-secrets'>   Optional: false
      LICENSE_PRIVATE_KEY:  <set to the key 'LICENSE_PRIVATE_KEY' in secret 'groupdocs-secrets'>  Optional: false
      ISOLATED_CONVERSION:  true

Please also check if value of ISOLATED_CONVERSION=true is correct

Screenshot 2026-07-22 at 4.25.36 PM.png (6.0 KB)

Hi, @piyush.rajput , please ensure you have using the image tagged as groupdocs/conversion-cloud:26.4.6 and you see the message “Isolated conversion mode enabled” right after “The license has been set.” in the log.
I have just pulled the image on a test host and it writes Isolated conversion mode in the log, so there is correct version at DockerHub.
Unless you don’t see the message in the log, we cannot be sure if isolated conversion there.

Hi @sergei.terentev ,
After deploying 26.4.6 instead of latest then I am seeing those logs:

info: GroupDocs.Conversion.Cloud.Web.Startup[0]
      The license has been set.
info: GroupDocs.Conversion.Cloud.Web.Startup[0]
      Isolated conversion mode enabled
Hosting environment: Production
Content root path: /app
Now listening on: http://[::]:80
Application started. Press Ctrl+C to shut down.

Ok, now the isolated mode works as expected. Please test it and monitor memory consumption changes. And, if issue still there, please collect and share logs (last 50-100 lines) of pod with OOMKilled status (previous instance).

Sure @sergei.terentev
Thank you

Hi @sergei.terentev ,
Issues seems resolved now after yesternight deployment around 8PM IST. Memory is getting released. And also getting logs in the pods.
Screenshot 2026-07-23 at 10.16.06 AM.png (200.6 KB)

Logs:

e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Web.Startup[0]
      The license has been set.
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Web.Startup[0]
      Isolated conversion mode enabled
Hosting environment: Production
Content root path: /app
Now listening on: http://[::]:80
Application started. Press Ctrl+C to shut down.
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #1 | pdf -> png | input 690 KB | avg input 690 KB | WorkingSet 333 MB (4% of 9216 MB) | GC heap 232 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #2 | pdf -> png | input 855 KB | avg input 772 KB | WorkingSet 139 MB (2% of 9216 MB) | GC heap 8 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #3 | pdf -> png | input 855 KB | avg input 800 KB | WorkingSet 139 MB (2% of 9216 MB) | GC heap 7 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #4 | pdf -> png | input 393 KB | avg input 698 KB | WorkingSet 141 MB (2% of 9216 MB) | GC heap 6 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #5 | pdf -> png | input 1188 KB | avg input 796 KB | WorkingSet 144 MB (2% of 9216 MB) | GC heap 9 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #6 | pdf -> png | input 599 KB | avg input 763 KB | WorkingSet 143 MB (2% of 9216 MB) | GC heap 7 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #7 | pdf -> png | input 599 KB | avg input 740 KB | WorkingSet 144 MB (2% of 9216 MB) | GC heap 8 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #8 | pdf -> png | input 65 KB | avg input 655 KB | WorkingSet 142 MB (2% of 9216 MB) | GC heap 5 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #9 | pdf -> png | input 503 KB | avg input 638 KB | WorkingSet 142 MB (2% of 9216 MB) | GC heap 6 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #10 | pdf -> png | input 503 KB | avg input 625 KB | WorkingSet 142 MB (2% of 9216 MB) | GC heap 6 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #11 | docx -> png | input 14 KB | avg input 569 KB | WorkingSet 141 MB (2% of 9216 MB) | GC heap 6 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #12 | docx -> pdf | input 14 KB | avg input 523 KB | WorkingSet 141 MB (2% of 9216 MB) | GC heap 5 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #13 | docx -> png | input 14 KB | avg input 484 KB | WorkingSet 141 MB (2% of 9216 MB) | GC heap 6 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #14 | docx -> pdf | input 14 KB | avg input 450 KB | WorkingSet 141 MB (2% of 9216 MB) | GC heap 5 MB
e[40me[32minfoe[39me[22me[49m: GroupDocs.Conversion.Cloud.Business.ConversionFacade[0]
      Conversion #15 | pdf -> png | input 2062 KB | avg input 558 KB | WorkingSet 147 MB (2% of 9216 MB) | GC heap 9 MB

Thank you for looking into it on priority.

Hi, @piyush.rajput, Great! Thank you too, your feedback is really help to improve our products!

Hi @sergei.terentev
Any plan to release the ARM64 arch image ?

Hi, @piyush.rajput , yes, we can publish ARM64 image, if you need it, we already have it and tested it. However there are some issues with converting CAD file formats. We can publish it next week at DockerHub

1 Like

Hi, @piyush.rajput , we have updated the images at DockerHub (latest and 26.4.7). ARM64 Arch added there (amd64 and arm64 now supported). The Isolated conversion mode is on by default (no need to set it on in this version).

1 Like

Thank you @sergei.terentev for the update.

You Welcome!