- 02 Aug, 2017 12 commits
-
-
Tim Shaw authored
Add translation code for the RPCs as well. Bug 3956.
-
Morris Jette authored
-
Morris Jette authored
Add pack_job_id and pack_job_offset to accounting database. Modified sacct to accept pack job ID specification using "#+#" notation. Modified sstat to accept pack job ID specification using "#+#" notation.
-
Morris Jette authored
-
Tim Wickberg authored
-
Dominik Bartkiewicz authored
NULL is returned if the token is not found, testing against '\0' is wrong (although does work okay in older compilers). Fixes new GCC 7.1 warning.
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
was matching more than expected.
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Danny Auble authored
error messages that aren't really errors. Related to Bug 3997
-
- 01 Aug, 2017 11 commits
-
-
Morris Jette authored
Without this change, each component would generate separate email at job begin, end, etc.
-
Morris Jette authored
-
Morris Jette authored
If the pack job allocation partially failed, properly handle accounting and deallocation of burst buffer. Note, this should rarely happen.
-
Tim Shaw authored
Bug 3999
-
Tim Shaw authored
Default to 1, unless set to 0. Allow to be set to 0 even if GroupUpdateTime was not set before. Move down to alphabetical position in read_config.c as well. Bug 3956.
-
Dominik Bartkiewicz authored
Fix bug in selection of GRES bound to specific CPUs where the GRES count is 2 or more. Previous logic could allocate CPUs not available to the job. bug 4029
-
Tim Wickberg authored
-
Dominik Bartkiewicz authored
NULL is returned if the token is not found, testing against '\0' is wrong (although does work okay in older compilers). Fixes new GCC 7.1 warning.
-
Morris Jette authored
Highlight the need to use Slurm abstract CPU ID
-
Alejandro Sanchez authored
-
Alejandro Sanchez authored
When the node isn't actually rebooted, the BootTime isn't updated and Slurm doesn't consider that the node is returned to service, even if slurmd is responding. Bug 4039
-
- 31 Jul, 2017 1 commit
-
-
Tim Shaw authored
This will be fixed before 17.11, but is being left as-is on 17.02. Bug 3956.
-
- 29 Jul, 2017 1 commit
-
-
Morris Jette authored
-
- 28 Jul, 2017 12 commits
-
-
Danny Auble authored
to have 'socket=' in AuthInfo to work. This is to make it so people don't have to update their slurmdbd.conf's when upgrading (and to match documentation). Continuation of last commit Bug 4009
-
Danny Auble authored
connection. Bug 4009
-
Morris Jette authored
-
Morris Jette authored
-
Alejandro Sanchez authored
jobcomp/elasticsearch saves/load the state to/from elasticsearch_state. Since the jobcomp API isn't designed with save/load state operations, the plugin _save_state() isn't extern and not available from outside the plugin itself, thus it is highly coupled to fini() function. This state doesn't follow the same execution path as the rest of Slurm states, where in save_all_sate() they are all independently scheduled. So we save it manually here on a RPC of type REQUEST_CONTROL. This enables that when the Primary ctld issues a REQUEST_CONTROL to the Backup which is currently in controller mode, the Backup will save the state and when the Primary assumes control again it can process the saved pending jobs. The other way around was already controlled, because when the Primary is running in controller mode and the Backup issues a REQUEST_CONTROL, the Primary is shutdown and when breaking the ctld main() function while(1) loop, there was already a g_slurm_jobcomp_fini() call in place. Bug 3908
-
Morris Jette authored
-
Morris Jette authored
If a pack job is only partitially allocated resources (likely due due to limits), deallocate resources from those components which have been started and requeue them.
-
Dominik Bartkiewicz authored
-
Morris Jette authored
Previous logic could corrupt a pack job's tres_alloc_count array resulting in slurmctld's accounting module abort.
-
Dominik Bartkiewicz authored
Bug 3973.
-
Morris Jette authored
-
Morris Jette authored
Perform limit check on heterogeneous job as a whole at submit time to reject jobs that will never be able to run. Accepting pack jobs that can never start will have a significant effect on scheduling in general (blocking the queue).
-
- 27 Jul, 2017 3 commits
-
-
Morris Jette authored
Prevent the possibility (never observed) of the ping count going negative
-
Morris Jette authored
-
Alejandro Sanchez authored
When more than 1 ping cycle is spawned simultaneously (for instance REQUEST_PING + REQUEST_NODE_REGISTRATION_STATUS for the selected nodes), we do not track a separate ping_start time for each cycle. When ping_begin() is called, the information about the previous ping cycle is lost. Then when ping_end() is called for the first of the two cycles, we set ping_start=0, which is incorrectly used to see if the last cycle ran for more than PING_TIMEOUT seconds (100s), thus incorrectly triggering the: error("Node ping apparently hung, many nodes may be DOWN or configured " "SlurmdTimeout should be increased"); Bug 3914
-