bcdb rebuild is not rebuilding the index correctly

[serboctor]

Steps:

• start the explorer. j.servers.threebot.local_start_explorer(background=True)
• create any object. I use the farm actors as an example:

  cl = j.clients.threebot.explorer.actors_get("tfgrid.directory")
  farm = {
      "threebot_id": 1,
      "name": f"stresstest{j.data.idgenerator.generateXCharID(5)}",
      "email": "farm@tf.com",
      "wallet_addresses": ["address"],
  }
  farm_id = cl.farms.register(farm).farm_id

• stop threebot server, but keep zdb running.
• get the farm model and list the farms, it lists fine.

  
  > bcdb = j.data.bcdb.instances["threebot_ZDB_tfgrid"]                                                                                                                                                    
  JSX> bcdb.storclient                                                                                                                                                                                        
  # JSXOBJ:
  ## jumpscale.zdb.client.1
  ID: 92
  - name                : threebot_ZDB_tfgrid
  - addr                : 127.0.0.1
  - port                : 9,900
  - secret_             : c2ace231021d72fb0329f0f7fb564a4a
  - nsname              : tfgrid
  - admin               : False
  - mode                : SEQ

JSX> farm_model = bcdb.model_get(url="tfgrid.directory.farm.1")
model:tfgrid.directory.farm.1
JSX> farm_model.find()
[## tfgrid.directory.farm.1 ID: 1

• threebot_id : 1

• iyo_organization :

• name : stresstestknzo7

• wallet_addresses :

  • address

    tfgrid.directory.location.1

    • city :
    • country :
    • continent :
    • latitude : 0
    • longitude : 0

• email : farm@tf.com

• resource_prices : []

• prefix_zero : ::/128 ]

  - delete the farm using the storclient directly, not through object deletion. The aim is to have inconsistent index.

  bcdb.storclient.delete(farm_id)

  
  - list the farms again, it will crash because the id is sitll in the index but removed from zdb. This should be fixed by rebuilding the index.

• exit kosmos and run bcdb rebuild in the shell.
• open the shell, and try to list the farms again, you'll still find the error.

It worked for me only when I deleted the sql index file and rebuild again. Also, @andrewayoub couldnt reproduce it with the rebuild_index function on the bcdb instance. It happens only with the bcdb rebuild command.

[andrewayoub]

when you rum bcdb rebuild it starts a new process and since bcdb models are lazy loaded, it will only be loaded when used, that's why I had to get all objects in order to enforce loading all models before rebuilding.

https://github.com/threefoldtech/jumpscaleX_core/commit/3a67d530e16766db7695d0e3288308b4e2be156d

[Dina-Abd-Elrahman]

verified