locked
Any issues deleting then recreating same index? RRS feed

  • Question

  • Hi

     To delete all the documents from a large index, we are deleting and then recreating the index ... and then reloading the data straight away.

     Is this likely to cause problems?

     We notice part way through the reload we the client connection times out. We wonder if there needs to be a delay between deleting the index and reloading so that Azure Search can eliminate the old index. 

    Thanks

    Thursday, January 7, 2016 5:54 PM

Answers

  • Followed up offline with Mark.

    This is related to this thread. In short, it is the result of a known issue in Azure Search, not high indexing load. See the other thread for details.

    Thursday, January 14, 2016 10:46 PM
    Moderator

All replies

  • If you're getting connection timeouts, it means your indexing load is too heavy for your service to handle. There are a few things you can do to improve this:

    1. Increase the density of documents per batch that you send.
    2. Decrease the rate at which you send indexing requests.
    3. Decrease the degree of parallelism if you're sending multiple batches concurrently.
    4. Scale up the number of partitions in your service. This increases write throughput.

    In general it would be better to avoid completely re-populating your index often. Can you describe your scenario in more detail? There may be better ways to keep your index up-to-date, depending on where your data is sourced.

    -Bruce

    Thursday, January 7, 2016 7:14 PM
    Moderator
  • Followed up offline with Mark.

    This is related to this thread. In short, it is the result of a known issue in Azure Search, not high indexing load. See the other thread for details.

    Thursday, January 14, 2016 10:46 PM
    Moderator