Feature #8270
Facility to document creation of branch names and branch IDs
Description
Whenever RootInput
files are read in to an art
job, a check is made to ensure consistency in the BranchIDLists
between files. The BranchID
values contained on these lists are checksums that have been generated from the branch name. Although the branch name is not generally stored in the file, it is stored in the ProductRegistry
for products that have been kept for that process. Unfortunately this is not sufficient for debugging branch mismatch errors. Users occasionally see errors that a branch mismatch exists between input files, but then receive no further information. The BranchID
values are stored for the input files, but unless the mismatch corresponds to a product in the ProductRegistry
(i.e. is a kept product), there is no way to retrieve the name of the branch(es) that caused the branch mismatch.
The art team should add debugging information so that it is easier to resolve this kind of error. I envisage two options:
- Simply adding
mf::LogDebug
(or equivalent) calls whenever the checksum is made, so that the (BranchID
, branch name) pair is printed out, or - Add a dedicated
ProductTracker
service that tracks where each product is created, and possibly consumed (to anticipate a future development ofart
).
Associated revisions
History
#1 Updated by Christopher Green almost 6 years ago
- Category set to Metadata
- Assignee set to Kyle Knoepfel
- Target version set to 1.14.00
- Estimated time set to 1.00 h
- SSI Package art added
- SSI Package deleted (
)
We can easily implement (1). (2) is reasonable, but is best tackled when we implement a planned major overhaul of the metadata system.
#2 Updated by Kyle Knoepfel almost 6 years ago
- Status changed from New to Assigned
#3 Updated by Kyle Knoepfel almost 6 years ago
- Status changed from Assigned to Resolved
- % Done changed from 0 to 100
Whenever there is a branch mismatch, a printout now exists that looks like:
Cannot merge file '<some_rootfile>.root' due to a branch mismatch. Previous File File to merge ============================== 2556661413 2556661413 701045456 3413590908 3413590908 3893403090 3893403090 741045478 741045478 4113287237 4113287237 3717081787 3717081787 3129636888 3129636888 2464441433 2464441433 2344735897 2344735897 3825369613 3825369613 680204693 680204693 71657639 71657639 393939105 393939105 3944962766 3944962766 360663258 360663258 1023633757 1023633757 1832547360 1832547360 653380120 653380120 808754512 808754512 1079679939 1079679939 481487918 481487918 333880152 333880152 2511341726 2511341726 765148699 765148699 378294740 378294740 3611458318 3611458318 The BranchIDs above correspond to products that were created whenever the current input files were produced. The lists above must be identical. To determine which products these BranchIDs correspond to, rerun the process that produced the input files, enabling full debug output for the message service. Then 'grep' the log file for messages with 'BranchID' Contact the framework group for assistance.
Whenever a user then enables debug printout on the process, each registered product emits a message like:
%MSG-d BranchID: TriggerResultInserter:TriggerResults@Construction 15-Apr-2015 15:16:20 CDT ModuleConstruction Product created with branch id: [1838029815] from branch name: "art::TriggerResults_TriggerResults__MemoryTracker." %MSG
Where not only is the mapping given between the branch id and the branch name, but the location of the produces<...>()
call is also given: in the constructor of the TriggerResultInserter
module, with the module label TriggerResults
.
Implemented with b365da3dab0af967a5a7b4c44e4118d3251341a2.
#4 Updated by Kyle Knoepfel over 5 years ago
- Status changed from Resolved to Closed
Implement issue #8270