1. Created 9e93e936ebe5872c59b48bb9f4000729 at 1-967a00dff5e02add41819138abb3284d .

2. Failed write attempt to 9e93e936ebe5872c59b48bb9f4000729.

3. Updated 9e93e936ebe5872c59b48bb9f4000729 to rev 2-8a0e060e92a3e9ea8f5f11bd0249ec18.

4. Created 9e93e936ebe5872c59b48bb9f4000e8a and 9e93e936ebe5872c59b48bb9f4001b21.

5. Stopped.

1. Created one document, 50b43230770fec2b1a3ea794f20007fc.

2. Failed write attempt to 50b43230770fec2b1a3ea794f2001423 (which stopped writing the proper header).

3. Created two other documents, 50b43230770fec2b1a3ea794f2000ab5, 50b43230770fec2b1a3ea794f2001423.

4. Updated 50b43230770fec2b1a3ea794f2001423 with rev 2-ae3fdc78a21f25eeee4f4ca2970d3424.

5. Restarted.

6. Updated 50b43230770fec2b1a3ea794f20007fc to rev 2-372471c7b896e112d57cffccb7bfd41b.

7. Failed write attempt to 50b43230770fec2b1a3ea794f20007fc (which stopped writing the proper header).

8. Updated 50b43230770fec2b1a3ea794f20007fc to rev 3-72041bcdea70a0d76d4ddb343fdf23e7.

9. Restarted.

10. Updated 50b43230770fec2b1a3ea794f20007fc to rev 3-a4d9ab5f51593f312ab58ed49bf1de58.

11. Stopped.

To be recovered:

50b43230770fec2b1a3ea794f2000ab5 at rev 1- and 50b43230770fec2b1a3ea794f2001423 at revision 2-ae3fdc78a21f25eeee4f4ca2970d3424.

50b43230770fec2b1a3ea794f20007fc rev 3-72041bcdea70a0d76d4ddb343fdf23e7.

This database has one document with hundreds of conflicting updates for it's second revision all committed between restarts.

The updates to be recovered are contained in the multi-conflict-changes-log file.

One document was created and then an invalid write attempted, putting the db in a bad state.

Then test/test_writes.js was run for 200 seconds. At the end of the run this was the state of the db.

  {
    db_name: "testwritesdb"
    doc_count: 124969
    doc_del_count: 0
    update_seq: 124969
    purge_seq: 0
    compact_running: false
    disk_size: 54857478
    instance_start_time: "1281384140058211"
    disk_format_version: 5
  }

This was the state after restart:

  {
    db_name: "testwritesdb"
    doc_count: 1
    doc_del_count: 0
    update_seq: 1
    purge_seq: 0
    compact_running: false
    disk_size: 54857478
    instance_start_time: "1281384593876026"
    disk_format_version: 5
  }

Copy of multi_conflict. Then make some simple doc changes, added multi_conflict as an attachment to two documents. Commited, restarted, this is the state:

{ db_name: "multi_conflict_with_attach" doc_count: 3 doc_del_count: 0 update_seq: 7 purge_seq: 0 compact_running: false disk_size: 159833 instance_start_time: "1281462427760706" disk_format_version: 5 }

But in to conflict state and multi-conflict is added one last time to the third document. State before restart:

{ db_name: "multi_conflict_with_attach" doc_count: 3 doc_del_count: 0 update_seq: 8 purge_seq: 0 compact_running: false disk_size: 225554 instance_start_time: "1281462644703945" disk_format_version: 5 }

Recovery should find 200 previous docs + 1 attachment and a couple soft updates to docs.

Just like multi_conflict except there is a document write and an _ensure_full_commit before putting the db in a state the loses the updates.