- 02 Aug, 2018 17 commits
-
-
Danny Auble authored
-
Danny Auble authored
be lost and would remain pending with no way to continue. This causes issues when dealing with TRES since the first message going out would be the registration message to get said TRES. Since this never gets sent any step start would segfault on the matter.
-
Danny Auble authored
Bug 5094
-
Thomas Cadeau authored
-
Thomas Cadeau authored
Bug 5094
-
Thomas Cadeau authored
Bug 5094
-
Tim Wickberg authored
-
Brian Christiansen authored
-
Morris Jette authored
-
Morris Jette authored
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
to dictate what state of node is after reboot.
-
Brian Christiansen authored
update_node_msg_t->node_state is uint32_t but we were putting a uint16_t into it. NO_VAL16 wouldn't have gone up to the controller.
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
-
- 01 Aug, 2018 4 commits
-
-
Marshall Garey authored
Bug 5196.
-
Marshall Garey authored
Needed for node_record_count. Bug 5196.
-
Tim Wickberg authored
This reverts commit a10454f7.
-
Marshall Garey authored
Bug 5196
-
- 31 Jul, 2018 2 commits
-
-
Danny Auble authored
-
Morris Jette authored
Bug 5070
-
- 30 Jul, 2018 2 commits
-
-
Tim Wickberg authored
-
Tim Wickberg authored
-
- 28 Jul, 2018 6 commits
-
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
- 27 Jul, 2018 9 commits
-
-
Danny Auble authored
Bug 5468 This is a backport of commit cefc9ec1.
-
Danny Auble authored
-
Felip Moll authored
Bug 4918
-
Tim Wickberg authored
Collapse error message on to one line when they weren't previously, and fixup argument indentation as well while here. find . -name '*\.[ch]' -exec sed -i s/error\(\[\ \]*\"sched:\ /sched_error\(\"/ {} \;
-
Tim Wickberg authored
These calls will replace all these: error("sched: message...") with: sched_error("message...") This allows the _log_msg function to stop calling xstrncmp() on every log message while holding log_lock when SchedLogLevel > 0. This behavior has effectively removed all concurrency from slurmctld when SchedLogLevel is enabled, and prevented larger scale systems from being able to start up - see bug 3746 for an example of this.
-
Tim Wickberg authored
Update all calling locations, and as this is a static function change the name to _log_msg from log_msg while here.
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Dominik Bartkiewicz authored
scheduling cycle. Primarily caused by EnforcePartLimits=ALL. Bug 5452
-