{
  "id": 126896,
  "title": "Kaggle Bug?",
  "url": "/competitions/deepfake-detection-challenge/discussion/126896",
  "author_name": "",
  "post_date": "2020-01-21T03:18:06.897959300Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>When I was running kaggle kernel interactively, the RAM usage wasn't even close to 12 GB(just 2.7GB) and the kernel restarted and gave me a warning saying: used more RAM than avaiable.</p>",
  "messages": [
    {
      "id": "724317",
      "postDate": "01/21/2020 03:18:06",
      "content": "<p>When I was running kaggle kernel interactively, the RAM usage wasn't even close to 12 GB(just 2.7GB) and the kernel restarted and gave me a warning saying: used more RAM than avaiable.</p>",
      "rawMarkdown": "When I was running kaggle kernel interactively, the RAM usage wasn't even close to 12 GB(just 2.7GB) and the kernel restarted and gave me a warning saying: used more RAM than avaiable.",
      "votes": null
    },
    {
      "id": "725118",
      "postDate": "01/21/2020 20:32:09",
      "content": "<p>The RAM usage is sampled every 10s, so it's possible that the kernel dies with OOM before actually sampling the high memory which triggered the OOM.</p>\n\n<p>Also note that when consuming lots of RAM, the sample could also be delayed enough to be longer than 10s.</p>",
      "rawMarkdown": "The RAM usage is sampled every 10s, so it's possible that the kernel dies with OOM before actually sampling the high memory which triggered the OOM.\n\nAlso note that when consuming lots of RAM, the sample could also be delayed enough to be longer than 10s.",
      "votes": null
    },
    {
      "id": "725123",
      "postDate": "01/21/2020 20:33:35",
      "content": "<p>Hi,</p>\n\n<p>Did you get the current RAM usage reading (2.7Gb) from the editor UI?</p>\n\n<p>We had an issue yesterday where the RAM usage wasn't properly propagated to the notebook editor UI. The issue has been fixed now. This means that you may have been using more RAM than shown.</p>\n\n<p>Can you try again and tell me if you are still experiencing this issue?</p>\n\n<p>Thank you</p>",
      "rawMarkdown": "Hi,\n\nDid you get the current RAM usage reading (2.7Gb) from the editor UI?\n\nWe had an issue yesterday where the RAM usage wasn't properly propagated to the notebook editor UI. The issue has been fixed now. This means that you may have been using more RAM than shown.\n\nCan you try again and tell me if you are still experiencing this issue?\n\nThank you",
      "votes": null
    },
    {
      "id": "725232",
      "postDate": "01/21/2020 23:10:16",
      "content": "<p>Thanks. It is fixed now!</p>",
      "rawMarkdown": "Thanks. It is fixed now!",
      "votes": null
    },
    {
      "id": "725887",
      "postDate": "01/22/2020 15:22:00",
      "content": "<p>Hi Vicent, regarding the editor UI I was experiencing another bug (not sure if it's fixed now). For me, the RAM usage in the editor was always increasing like a memory leak. But if I ran the same code on my local computer, the RAM would be very stable. Furthermore, I submitted the kernel and it worked, which is surprising considering that if the leak were indeed happening it would definitively run OOM for the much bigger private dataset. </p>\n\n<p>Could this issue be the same bug you are mentioning? \nThank you</p>",
      "rawMarkdown": "Hi Vicent, regarding the editor UI I was experiencing another bug (not sure if it's fixed now). For me, the RAM usage in the editor was always increasing like a memory leak. But if I ran the same code on my local computer, the RAM would be very stable. Furthermore, I submitted the kernel and it worked, which is surprising considering that if the leak were indeed happening it would definitively run OOM for the much bigger private dataset. \n\nCould this issue be the same bug you are mentioning? \nThank you",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 725118,
      "author_name": "herbison",
      "author_url": "",
      "post_date": "01/21/2020 20:32:09",
      "content": "<p>The RAM usage is sampled every 10s, so it's possible that the kernel dies with OOM before actually sampling the high memory which triggered the OOM.</p>\n\n<p>Also note that when consuming lots of RAM, the sample could also be delayed enough to be longer than 10s.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 725123,
      "author_name": "rosebv",
      "author_url": "",
      "post_date": "01/21/2020 20:33:35",
      "content": "<p>Hi,</p>\n\n<p>Did you get the current RAM usage reading (2.7Gb) from the editor UI?</p>\n\n<p>We had an issue yesterday where the RAM usage wasn't properly propagated to the notebook editor UI. The issue has been fixed now. This means that you may have been using more RAM than shown.</p>\n\n<p>Can you try again and tell me if you are still experiencing this issue?</p>\n\n<p>Thank you</p>",
      "votes": null,
      "replies": [
        {
          "id": 725232,
          "author_name": "unkownhihi",
          "author_url": "",
          "post_date": "01/21/2020 23:10:16",
          "content": "<p>Thanks. It is fixed now!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 725887,
          "author_name": "arc144",
          "author_url": "",
          "post_date": "01/22/2020 15:22:00",
          "content": "<p>Hi Vicent, regarding the editor UI I was experiencing another bug (not sure if it's fixed now). For me, the RAM usage in the editor was always increasing like a memory leak. But if I ran the same code on my local computer, the RAM would be very stable. Furthermore, I submitted the kernel and it worked, which is surprising considering that if the leak were indeed happening it would definitively run OOM for the much bigger private dataset. </p>\n\n<p>Could this issue be the same bug you are mentioning? \nThank you</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "724317": "When I was running kaggle kernel interactively, the RAM usage wasn't even close to 12 GB(just 2.7GB) and the kernel restarted and gave me a warning saying: used more RAM than avaiable.",
    "725118": "The RAM usage is sampled every 10s, so it's possible that the kernel dies with OOM before actually sampling the high memory which triggered the OOM.\n\nAlso note that when consuming lots of RAM, the sample could also be delayed enough to be longer than 10s.",
    "725123": "Hi,\n\nDid you get the current RAM usage reading (2.7Gb) from the editor UI?\n\nWe had an issue yesterday where the RAM usage wasn't properly propagated to the notebook editor UI. The issue has been fixed now. This means that you may have been using more RAM than shown.\n\nCan you try again and tell me if you are still experiencing this issue?\n\nThank you",
    "725232": "Thanks. It is fixed now!",
    "725887": "Hi Vicent, regarding the editor UI I was experiencing another bug (not sure if it's fixed now). For me, the RAM usage in the editor was always increasing like a memory leak. But if I ran the same code on my local computer, the RAM would be very stable. Furthermore, I submitted the kernel and it worked, which is surprising considering that if the leak were indeed happening it would definitively run OOM for the much bigger private dataset. \n\nCould this issue be the same bug you are mentioning? \nThank you"
  },
  "source": "meta"
}