- 08 Sep, 2016 7 commits
-
-
Brian Christiansen authored
-
Brian Christiansen authored
Grab federations from db at startup instead of waiting for db_update and load from state if the db is down.
-
Brian Christiansen authored
-
Brian Christiansen authored
Instead of making a separate copy. All of the cluster_recs are now in the federation_rec with a pointer to the local cluster rec.
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
Select federations base off which clusters belong to them.
-
- 07 Sep, 2016 30 commits
-
-
Morris Jette authored
-
Morris Jette authored
Preserve node "RESERVATION" state when one of multiple overlapping reservations ends. Previous logic would clear the node's RESERVATION state flag when any one of the reservations on the node ended rather than keeping the node in RESERVATION state until the last reservation ended. bug 3057
-
Morris Jette authored
The logic is now heavier weight, so increase interval between tests from 2 to 5 seconds
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
Instead of making copies just use the pointers and stay in the read locks.
-
Brian Christiansen authored
-
Brian Christiansen authored
Before fixing fed.flags to be initialized to FEDERATION_NOT_SET, the federation's flags were being set to 0 when a federation modification happened. These tests verify that the flags stay after a federation modification.
-
Brian Christiansen authored
Was modeled after cluster adding where commit the check was being done before sending the changes to the dbd. But since the dbd isn't making any additional tables for federations -- like it does for clusters -- it can send to the dbd first and make sure they worked and then ask to commit the changes.
-
Brian Christiansen authored
Ran into the issue after initializing fed flags to NOTSET. Fed flags were always being set to 0 on modification.
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
Changed READ_LOCK to WRITE_LOCK since we are updating the node's select info (select_g_select_nodeinfo_set_all).
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
Inside locks.
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
Also fixes setting weight at cluster creation.
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
Don't store local cluster info in siblings. Store separately for state_save and "scontrol show fed".
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Morris Jette authored
This is required in order to make sure that a compute node reboots as requested; even if the slurmctld daemon restarts. Previously if the slurmctld daemon restarted, we couldn't tell if the compute node rebooted before or after the request. This commit expands upon the work of commit 4517c454 by adding the reboot request time to the node record, which is saved and restored by slurmctld. bug 3042
-
Morris Jette authored
-
Morris Jette authored
Handle case when slurmctld daemon restart while compute node reboot in progress. Return node to service rather than setting DOWN. bug 3042
-
- 06 Sep, 2016 3 commits
-
-
Morris Jette authored
-
Morris Jette authored
Add salloc_wait_nodes option to the SchedulerParameters parameter in the slurm.conf file controlling when the salloc command returns in relation to when nodes are ready for use (i.e. booted). bug 3043
-
Morris Jette authored
-