Collections as regular Arvados objects » History » Version 5
Peter Amstutz, 08/08/2014 09:46 AM
1 | 1 | Peter Amstutz | h1. Collections as regular Arvados objects |
---|---|---|---|
2 | |||
3 | * Collections are assigned regular UUIDs |
||
4 | 5 | Peter Amstutz | * Collections have a regular @owner_uuid@ instead |
5 | 1 | Peter Amstutz | * Manifest hash goes into the @portable_data_hash@ field |
6 | 5 | Peter Amstutz | * @manifest_text@ and @portable_data_hash@ are immutable |
7 | 3 | Peter Amstutz | * Add mutable @name@, @description@, @properties@ fields |
8 | * Continue to use manifest text hash (@portable_data_hash@) for crunch |
||
9 | ** Choosers set the value to the manifest hash, record the selected collection uuid on link_uuid |
||
10 | ** Task output is manifest fragment or manifest hash+reader token stored in Keep |
||
11 | ** Crunch-job continues to set job output to manifest hash |
||
12 | ** Add optional @output_name@ and @output_owner@ fields to job submission |
||
13 | ** When job completes, collection is created under @output_owner@, given @output_name@, and the collection uuid is recorded under new @output_uuid@ field |
||
14 | 1 | Peter Amstutz | |
15 | 4 | Peter Amstutz | (Side note: job should not be marked as successful/completed until all its output and logs have been uploaded, see #3164) |
16 | 3 | Peter Amstutz | |
17 | h2. Migration |
||
18 | |||
19 | 2 | Peter Amstutz | * Create collection objects from name links |