embedded documents update for array elements lead to postgres restart

[zhang699]

I have use pgAdmin GUI to test embedded update by mongo_fdw

Tested Documents

collection: warehouse
{
  _id: ObjectId("61d7ad616372f5550c92ac4b"),
  elements: ["a", "b"]
}

SQL Schema

id: name
elements: text[]

• Try the query

  UPDATE warehouse SET elements[1] = 'test' WHERE _id = '61d7ad616372f5550c92ac4b'

• Query Log

  Connection to the server has been lost

• Postgres Console log

  postgres_1  | /app/mongo_fdw/mongo-c-driver/src/libbson/src/bson/bson.c:326 _bson_append_va(): precondition failed: !(bson->flags & BSON_FLAG_IN_CHILD)
  postgres_1  | 2022-01-11 08:01:52.535 UTC [1] LOG:  server process (PID 1346) was terminated by signal 6: Aborted
  postgres_1  | 2022-01-11 08:01:52.535 UTC [1] DETAIL:  Failed process was running: UPDATE warehouse SET elements[1] = 'test' WHERE _id = '61d7ad616372f5550c92ac4b'
  postgres_1  | 2022-01-11 08:01:52.535 UTC [1] LOG:  terminating any other active server processes
  mongo_1     | {"t":{"$date":"2022-01-11T08:01:52.534+00:00"},"s":"I",  "c":"NETWORK",  "id":22944,   "ctx":"conn15","msg":"Connection ended","attr":{"remote":"192.168.208.4:43870","connectionId":15,"connectionCount":10}}
  postgres_1  | 2022-01-11 08:01:52.587 UTC [1] LOG:  all server processes terminated; reinitializing
  postgres_1  | 2022-01-11 08:01:52.754 UTC [1348] LOG:  database system was interrupted; last known up at 2022-01-11 07:31:27 UTC
  postgres_1  | 2022-01-11 08:01:52.755 UTC [1350] FATAL:  the database system is in recovery mode
  postgres_1  | 2022-01-11 08:01:52.756 UTC [1349] FATAL:  the database system is in recovery mode
  postgres_1  | 2022-01-11 08:01:52.915 UTC [1348] LOG:  database system was not properly shut down; automatic recovery in progress
  postgres_1  | 2022-01-11 08:01:52.918 UTC [1348] LOG:  redo starts at 0/1726960
  postgres_1  | 2022-01-11 08:01:52.918 UTC [1348] LOG:  invalid record length at 0/1726A48: wanted 24, got 0
  postgres_1  | 2022-01-11 08:01:52.918 UTC [1348] LOG:  redo done at 0/1726A10 system usage: CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s

[vaibhavdalvi93]

Thanks, @zhang699 for reporting an issue.

I can reproduce the same issue at my end. This is a bug.

The root cause of the issue is that while preparing the query document, provided the wrong first argument to the function BsonAppendUTF8() which is called by function AppendMongoValue().

We will work on this issue in near future.

[zhang699]

Thanks, @zhang699 for reporting an issue.

I can reproduce the same issue at my end. This is a bug.

The root cause of the issue is that while preparing the query document, provided the wrong first argument to the function BsonAppendUTF8() which is called by function AppendMongoValue().

We will work on this issue in near future.

Great !!

[vaibhavdalvi93]

Hi @zhang699 ,

The reported issue has been resolved. It would be good if you please confirm from your side on latest head/release and close the ticket if it's been resolved.