{
  "id": 680699,
  "title": "Long Submission Queue and Unexpected Timeouts",
  "url": "/competitions/deep-past-initiative-machine-translation/discussion/680699",
  "author_name": "",
  "post_date": "2026-03-10T10:36:44.889684100Z",
  "votes": 9,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Recently I have noticed that Kaggle submission processing times have increased significantly. Previously, I was able to receive results in around 2 hours, but now it often takes 4–5 hours or even longer.</p>\n<p>In addition, some of my submissions are ending with a timeout, even though they previously completed successfully within the allowed limits. This makes it difficult to iterate and test improvements during the competition.</p>\n<p>Is this related to increased platform load or any recent infrastructure changes? If there are any recommended adjustments (e.g., resource limits or submission strategies), I would appreciate the guidance.</p>",
  "messages": [
    {
      "id": "3419286",
      "postDate": "03/10/2026 10:36:44",
      "content": "<p>Recently I have noticed that Kaggle submission processing times have increased significantly. Previously, I was able to receive results in around 2 hours, but now it often takes 4–5 hours or even longer.</p>\n<p>In addition, some of my submissions are ending with a timeout, even though they previously completed successfully within the allowed limits. This makes it difficult to iterate and test improvements during the competition.</p>\n<p>Is this related to increased platform load or any recent infrastructure changes? If there are any recommended adjustments (e.g., resource limits or submission strategies), I would appreciate the guidance.</p>",
      "rawMarkdown": "Recently I have noticed that Kaggle submission processing times have increased significantly. Previously, I was able to receive results in around 2 hours, but now it often takes 4–5 hours or even longer.\n\nIn addition, some of my submissions are ending with a timeout, even though they previously completed successfully within the allowed limits. This makes it difficult to iterate and test improvements during the competition.\n\nIs this related to increased platform load or any recent infrastructure changes? If there are any recommended adjustments (e.g., resource limits or submission strategies), I would appreciate the guidance.",
      "votes": null
    },
    {
      "id": "3419357",
      "postDate": "03/10/2026 14:47:57",
      "content": "<p>It's the same thing I have faced many times here :(</p>",
      "rawMarkdown": "It's the same thing I have faced many times here :(",
      "votes": null
    },
    {
      "id": "3419365",
      "postDate": "03/10/2026 15:03:29",
      "content": "<p><a href=\"https://www.kaggle.com/herbison\" target=\"_blank\">@herbison</a> Hi it seems P100 is experience queue since yesterday, is something wrong or just more people are using it? Thanks</p>",
      "rawMarkdown": "herbison Hi it seems P100 is experience queue since yesterday, is something wrong or just more people are using it? Thanks",
      "votes": null
    },
    {
      "id": "3419371",
      "postDate": "03/10/2026 15:09:51",
      "content": "<p>We're maxed out on P100s and its an older accelerator so there's no more to get.\nThe options for now are: switch to T4s, wait for a P100.</p>\n<p>We'll work on getting more capacity overall in the fleet, but there won't be more P100s, so capacity will always be tight (queueing) at the daily peak usage.</p>",
      "rawMarkdown": "We're maxed out on P100s and its an older accelerator so there's no more to get.\nThe options for now are: switch to T4s, wait for a P100.\n\nWe'll work on getting more capacity overall in the fleet, but there won't be more P100s, so capacity will always be tight (queueing) at the daily peak usage.",
      "votes": null
    },
    {
      "id": "3419376",
      "postDate": "03/10/2026 15:14:52",
      "content": "<p>Thanks! I appreciate your quick (and detailed) reply! you are doing a phenomenal job</p>",
      "rawMarkdown": "Thanks! I appreciate your quick (and detailed) reply! you are doing a phenomenal job",
      "votes": null
    },
    {
      "id": "3421353",
      "postDate": "03/15/2026 10:48:55",
      "content": "<p>me too, always internet timeout</p>",
      "rawMarkdown": "me too, always internet timeout",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3419357,
      "author_name": "bhaweshsinha07",
      "author_url": "",
      "post_date": "03/10/2026 14:47:57",
      "content": "<p>It's the same thing I have faced many times here :(</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3419365,
      "author_name": "honganzhu",
      "author_url": "",
      "post_date": "03/10/2026 15:03:29",
      "content": "<p><a href=\"https://www.kaggle.com/herbison\" target=\"_blank\">@herbison</a> Hi it seems P100 is experience queue since yesterday, is something wrong or just more people are using it? Thanks</p>",
      "votes": null,
      "replies": [
        {
          "id": 3419371,
          "author_name": "herbison",
          "author_url": "",
          "post_date": "03/10/2026 15:09:51",
          "content": "<p>We're maxed out on P100s and its an older accelerator so there's no more to get.\nThe options for now are: switch to T4s, wait for a P100.</p>\n<p>We'll work on getting more capacity overall in the fleet, but there won't be more P100s, so capacity will always be tight (queueing) at the daily peak usage.</p>",
          "votes": null,
          "replies": [
            {
              "id": 3419376,
              "author_name": "honganzhu",
              "author_url": "",
              "post_date": "03/10/2026 15:14:52",
              "content": "<p>Thanks! I appreciate your quick (and detailed) reply! you are doing a phenomenal job</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 3421353,
      "author_name": "zhaoxinlong",
      "author_url": "",
      "post_date": "03/15/2026 10:48:55",
      "content": "<p>me too, always internet timeout</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3419286": "Recently I have noticed that Kaggle submission processing times have increased significantly. Previously, I was able to receive results in around 2 hours, but now it often takes 4–5 hours or even longer.\n\nIn addition, some of my submissions are ending with a timeout, even though they previously completed successfully within the allowed limits. This makes it difficult to iterate and test improvements during the competition.\n\nIs this related to increased platform load or any recent infrastructure changes? If there are any recommended adjustments (e.g., resource limits or submission strategies), I would appreciate the guidance.",
    "3419357": "It's the same thing I have faced many times here :(",
    "3419365": "herbison Hi it seems P100 is experience queue since yesterday, is something wrong or just more people are using it? Thanks",
    "3419371": "We're maxed out on P100s and its an older accelerator so there's no more to get.\nThe options for now are: switch to T4s, wait for a P100.\n\nWe'll work on getting more capacity overall in the fleet, but there won't be more P100s, so capacity will always be tight (queueing) at the daily peak usage.",
    "3419376": "Thanks! I appreciate your quick (and detailed) reply! you are doing a phenomenal job",
    "3421353": "me too, always internet timeout"
  },
  "source": "meta"
}