coll/base verbose, and neg priority cleanup #1834
Conversation
jjhursey
commented
Jun 30, 2016
jjhursey
commented
- Negative priority cleanup
- Fixes a cleanup segv in hcoll if it is asked to take a negative priority and has to cleanup.
- Improves coll/base verbose messages making it easier to see the set of collectives selected and being queried (as some might be rejected based on negative priority).
|
Build Failed with GNU compiler! Please review the log, and get in touch if you have questions. |
|
Build Failed with XL compiler! Please review the log, and get in touch if you have questions. |
|
(IBM Jenkins) Per #1833 the cluster seems to be having problems - I'm disabling the IBM tests for now while I diagnose. |
|
bot:retest |
|
Build Failed with XL compiler! Please review the log, and get in touch if you have questions. |
|
bot:ibm:retest |
1 similar comment
|
bot:ibm:retest |
|
:bot:retest |
|
Just testing our Jenkins setup: |
* Print a verbose message if the component was disqualified because of
a negative priority.
* If a disqualified component provided a module, release it.
* Display list of selected components in priority order
- During the process of volunteering collective functions for a
communicator, print the component name and priority. This will
cause the verbose messages to be displayed in reverse priority
order (lowest priority first, up to highest). This is helpful
when determining which collective components are active in which
order for a given communicator.
To see the messages you need the following MCA parameter set to 9
or higher: `-mca coll_base_verbose 9`
* Adjust verbose for commonly needed verbose output from 10 to 9 to
make it easier to access this information.
* If hcoll is given a negative priority, but not enabled=0 then the module is constructed, but then destructed before calling it's query(). So the previous pointers are not initialized. If we try to OBJ_RELEASE them in a debug build an assert will fire. This commit adds some protection against that and initializes the _module pointers to NULL.
jjhursey
force-pushed
the
topic/coll-base-verbose
branch
from
950b8a5 to
0a09f8b
Compare
|
Mellanox failure is unrelated. It is the: [jenkins01:09635] listen_thread: accept() failed: Invalid argument (22).failure that we have been seeing in a lot of other PRs - I think this is a problem with the master branch. |
|
@jjhursey Has an issue been filed on master about that listen_thread error? |
|
No, but it should be... |
4 participants
