You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I only noticed while testing new (extensice) changes to DagmanCreator and DagmanSubmitter that crab kill fails. as indicated in #8893 (comment) (BUG 5).
The bug was independent and present already in v3.250109 which we have in preprod since then and luckily did not deploy in production yet.
uses the SQL from Task.GetReadyTasks_sql which retrieves in the WHERE only a subset of the DB Task table columns. clusterid is not one of them !! In spite of what the comment in the API above states. 😠
Of course this must be because initially the tm_* columns where "all the information", then developers added more columns and not all SQL's have been changed. I do not know if it is possible to have code which retrieves all colums and properly names them, to avoid the need to list column names in so many places...
Anyhow, I do not feel like changing the REST code now. Will add a call in DagmanKiller to retrieve the full info about the task in order to check clusterid
The text was updated successfully, but these errors were encountered:
belforte
added a commit
to belforte/CRABServer
that referenced
this issue
Jan 29, 2025
I only noticed while testing new (extensice) changes to DagmanCreator and DagmanSubmitter that
crab kill
fails. as indicated in #8893 (comment) (BUG 5).The bug was independent and present already in v3.250109 which we have in preprod since then and luckily did not deploy in production yet.
We need to add
crab kill
to our validation.The problem is this line2
CRABServer/src/python/TaskWorker/Actions/DagmanKiller.py
Lines 42 to 44 in 5bdfcbf
introduced in 8ed2195 to fix #8874
The problem is that the
task
dictionary at this point in the code is what MasterWorker retrieved via a GET toworkflowdb
REST API inCRABServer/src/python/TaskWorker/MasterWorker.py
Line 430 in 5bdfcbf
and that API as per
CRABServer/src/python/CRABInterface/RESTWorkerWorkflow.py
Lines 86 to 89 in 5bdfcbf
uses the SQL from Task.GetReadyTasks_sql which retrieves in the
WHERE
only a subset of the DB Task table columns.clusterid
is not one of them !! In spite of what the comment in the API above states. 😠Of course this must be because initially the
tm_*
columns where "all the information", then developers added more columns and not all SQL's have been changed. I do not know if it is possible to have code which retrieves all colums and properly names them, to avoid the need to list column names in so many places...Anyhow, I do not feel like changing the REST code now. Will add a call in DagmanKiller to retrieve the full info about the task in order to check
clusterid
The text was updated successfully, but these errors were encountered: