- 05 Apr, 2019 1 commit
-
-
Alejandro Sanchez authored
Bug 6791.
-
- 03 Apr, 2019 2 commits
-
-
Alejandro Sanchez authored
This prevents rebuilding a job's dependency string when it has at least one invalid (never satisfied) dependency, no matter if such invalid dependency has already been purged (after MinJobAge) or not. This can be useful to track down the culprit invalid dependencies even after they are gone from ctld's in-memory job list. The flag is cleared upon a successful job dependency update or after another job in the dependency list has been satisfied if such list is composed with the '?' symbol (OR'ed). Bug 5851.
-
Alejandro Sanchez authored
Job dependencies separated by "?" (OR'ed) should make the dependant job be independent as soon as any of the dependencies are resolved to be satisfied. Without this patch, if an invalid (non satisfiable) dependency was resolved before a satisfiable one, then the dependant job would never become independent, even after the satisfiable one got eventually resolved. Bug 5851.
-
- 27 Mar, 2019 1 commit
-
-
Dominik Bartkiewicz authored
Bug 6750.
-
- 26 Mar, 2019 1 commit
-
-
Alejandro Sanchez authored
Bug 6710.
-
- 21 Mar, 2019 1 commit
-
-
Marshall Garey authored
Change to %pJ identifier while here and merge different partition priorities into a single log message line. Bug 6663.
-
- 20 Mar, 2019 5 commits
-
-
Alejandro Sanchez authored
Bug 6723
-
Albert Gil authored
Regression from enhancement 4506. Bug 6680
-
Danny Auble authored
Bug 6662
-
Alejandro Sanchez authored
Previously the state reason could remain as WAIT_NO_REASON even after backfill evaluation. This should improve system feedback/responsiveness user perception. Bug 6594.
-
Brian Christiansen authored
Continuation of 69d78159 Bug 6500
-
- 19 Mar, 2019 2 commits
-
-
Danny Auble authored
Issue was from ade9101e. The problem was an over simplified if statement which when ran multiple times would set the start time incorrectly. Bug 6697
-
Alejandro Sanchez authored
Even if main scheduler doesn't allocate resources for hetjobs, the queue list should be composed by all types of jobs. Otherwise, lower priority regular jobs could be allocated resources by main scheduler while higher priority hetjobs are waiting for a backfill cycle. Bug 6593.
-
- 15 Mar, 2019 1 commit
-
-
Matt Ezell authored
Bug 6679
-
- 13 Mar, 2019 1 commit
-
-
Alejandro Sanchez authored
Instruct the backfill scheduler to attempt to start a heterogeneous job as soon as all of its components are determined able to do so. Bug 5579.
-
- 12 Mar, 2019 1 commit
-
-
Dominik Bartkiewicz authored
There are at least two points where this can return a false positive error message which can be confusing for users. Continuation dc583bd1 Bug 6437
-
- 11 Mar, 2019 1 commit
-
-
Danny Auble authored
-
- 07 Mar, 2019 7 commits
-
-
Tim Wickberg authored
-
Danny Auble authored
Backport of 21b9a39b Bug 6625
-
Tim Wickberg authored
-
Nathan Rini authored
Print error: for each new line. Bug 6142
-
Nathan Rini authored
Bug 6142
-
Dominik Bartkiewicz authored
Bug 6584
-
Matt Ezell authored
response_cluster_rec stored a pointer to data in the controller config that gets freed when the controller is reconfigured. Bug 6584
-
- 06 Mar, 2019 1 commit
-
-
Nate Rini authored
This should never happen, but is a nice safeguard. Bug 6566
-
- 05 Mar, 2019 2 commits
-
-
Danny Auble authored
Bug 6612
-
Nathan Rini authored
All of the buffers are 24 characters, require the input string is less than 23 characters and null combined to be parsed. Remove regression added in b9699c2f Bug 6636.
-
- 01 Mar, 2019 2 commits
-
-
Broderick Gardner authored
What was happening in a maintenance reservation was every 5 seconds the slurmctld was sending a message to the slurmdbd telling it the node's down/drain state was in a maintenance state as well. We really only needed to know about this once. Bug 6487
-
Broderick Gardner authored
(DRAIN+MAINT+IDLE) Before it would just print MAINT Bug 6487
-
- 28 Feb, 2019 9 commits
-
-
Alejandro Sanchez authored
Bug 4296
-
Brian Christiansen authored
to avoid having to send alias list for cloud nodes that exist in DNS. This is especially important for large cloud environments (e.g. thousands of nodes) as the alias list environment will be too large for execve(). Bug 6589
-
Brian Christiansen authored
Bug 6587
-
Brian Christiansen authored
This is useful in a cloud environment where the nodes come and go out of DNS. Bug 6592
-
Marshall Garey authored
Bug 6519.
-
Brian Christiansen authored
Continuation of 324404de Bug 6433
-
Brian Christiansen authored
Continuation of c2cdde85 Bug 6433
-
Tim Wickberg authored
Same as salloc/sbatch --gres option. Bug 6582.
-
Marshall Garey authored
salloc --no-shell jobs aren't interactive, and won't respond to pings from the slurmctld, so they were previously getting killed after InactiveLimit seconds, even when they had a running step. InactiveLimit no longer applies for salloc --no-shell jobs. Bug 6560.
-
- 27 Feb, 2019 2 commits
-
-
Michael Hinton authored
Fixes smail to handle change in slurmctld/agent.c#mail_job_info() output introduced in commit 5f1cc8a8. seff has incomplete job array support, but if one happens to come through smail, at least let it print out what it can (i.e. the last array task in the job array). Bug 6536
-
Tim Wickberg authored
Allow sites to use hostname as given by libc gethostname() instead of the first domain used globally as hostname. Bug 6532. Co-authored-by:
Nate Rini <nate@schedmd.com>
-