Commits · 2836c84f717b69959f1bb5dbf3a2af28e981406c · Manuel G. Marciani / ces_slurm_simulator

22 Sep, 2003 1 commit
- Describe how inconsistencies are handled with respec to the --relative · 2836c84f
  Moe Jette authored Sep 21, 2003
```
and --nodes options.
```
  2836c84f
21 Sep, 2003 11 commits
- Print a warning message if the number of nodes remaining based upon the · 8d0bdacc
  Moe Jette authored Sep 21, 2003
```
--relative option is lower than the node count specified. The --relative
option takes precedence.
```
  8d0bdacc
- Record updates for 0.2.17 release. · 939a03c0
  Moe Jette authored Sep 21, 2003
  
  939a03c0
- Clean up error messages. · 38c62432
  Moe Jette authored Sep 21, 2003
  
  38c62432
- Clean up error message format. · 62ff4f15
  Moe Jette authored Sep 21, 2003
  
  62ff4f15
- Distinguish in logs between jobs that reach their job time limit and those · fccce86c
  Moe Jette authored Sep 21, 2003
```
that reach the slurm inactivity time limit.
```
  fccce86c
- Parameterize the maximum time for the backup controller to reliquish · f3bce2df
  Moe Jette authored Sep 21, 2003
```
control (it needs to complete all pending RPCs and save state before
the primary reads state and takes over).
```
  f3bce2df
- Restructure code so state is restored only when taking primary controller · 3c4c1e00
  Moe Jette authored Sep 21, 2003
```
responsibilities (backup was routinely reading at startup).
```
  3c4c1e00
- Minor code restructuring to improve responsiveness of slurmctld backup · ac67be19
  Moe Jette authored Sep 21, 2003
```
server to shutdown request.
```
  ac67be19
- Create _disable_signal function for clarity and add disable of SIGIO, · 1c6fafc5
  Moe Jette authored Sep 21, 2003
```
SIGPWR, and SIGLOST.
```
  1c6fafc5
- Fix bug that had non-responding DRAINING node go into state DRAINED · ce1d362d
  Moe Jette authored Sep 20, 2003
```
and when returned to service went improperly back into state DRAINING
(job counter was inconsistent).
```
  ce1d362d
- A non-responding node in DRAINING state will have its jobs killed and · b5cc2f0d
  Moe Jette authored Sep 20, 2003
```
transition to DRAINED state.
```
  b5cc2f0d
20 Sep, 2003 6 commits
- Don't bother to shutdown backup controller on reconfig request. Its · 987261ff
  Moe Jette authored Sep 20, 2003
```
data will not be used and the process is too slow anyway.
```
  987261ff
- Save all state when the last node associated with a job records its · 78fdfe9c
  Moe Jette authored Sep 20, 2003
```
EPILOG_COMPLETE_MESSAGE. At this time the job is COMPLETED and all
associated nodes available.
```
  78fdfe9c
- o NEWS updates for 0.2.17 · 1ecbd59f
  Mark Grondona authored Sep 19, 2003
  
  1ecbd59f
- Block SIGPIPE throughout slurmctld. · 9b1399cf
  Moe Jette authored Sep 19, 2003
  
  9b1399cf
- o add --relative, -r option to srun to allow running job steps · 74d08e4c
  Mark Grondona authored Sep 19, 2003
```
   on nodes relative to the current allocation.
 o srun no longer sends SIGKILL to job if one task is killed except
   if --no-allocate is used. (the job will otherwise be killed by
   the controller anyway)
```
  74d08e4c
- o reenable chkconfig on install · 99874548
  Mark Grondona authored Sep 19, 2003
  
  99874548
19 Sep, 2003 11 commits
- o update to latest hostlist.[ch] from LSD-Tools. Includes hostlist_nth() · af4213fe
  Mark Grondona authored Sep 19, 2003
```
   function.
```
  af4213fe
- o added code to clear any stale job step entries in shm if shared · 2f89ecd2
  Mark Grondona authored Sep 19, 2003
```
   memory appears to be full.
```
  2f89ecd2
- o add ECONNRESET to the set of errnos that close the IO object on · da5571b9
  Mark Grondona authored Sep 19, 2003
```
   write() call.
```
  da5571b9
- o fixes related to kill_job rpc: · 0dd2bfcd
  Mark Grondona authored Sep 19, 2003
```
  - instead of attempting to kill pending threads, immediately
    exit wait_for_procs if a thread is already waiting for job.
  - if wait_for_procs fails (thread already waiting), exit w/out
    sending epilog complete rpc.
```
  0dd2bfcd
- Change logging level of detailed communication errors from type info to · 019f229e
  Moe Jette authored Sep 19, 2003
```
type debug.
```
  019f229e
- Re-issue _slurm_close functions on EINTR to avoid leaving orphan file · 97bee5a0
  Moe Jette authored Sep 19, 2003
```
descriptors. This was needed in several grouped functions (e.g.
slurm_send_recv_rc_msg and slurm_send_only_node_msg, which combine
open, send, receive, and close functions for simplicity).
```
  97bee5a0
- Add a couple of xassert() calls. · e48ce3ac
  Moe Jette authored Sep 19, 2003
  
  e48ce3ac
- Change return in main to exit(0). · 75205e1a
  Moe Jette authored Sep 19, 2003
  
  75205e1a
- Decrease number of node re-registrations requests from the number of · 2c7f3570
  Moe Jette authored Sep 19, 2003
```
RPC server threads to that number - 2. This should slightly reduce the
incomming RPC load.
```
  2c7f3570
- Set stack size to 1MB per pthread. It was defaulting to 8MB per pthread · 27a4f044
  Moe Jette authored Sep 19, 2003
```
and may have been exhausting virtual memory, resulting in the death of
slurmctld.
```
  27a4f044
- o Also block EPIPE in srun (gnats:285) · d5b3731e
  Mark Grondona authored Sep 18, 2003
  
  d5b3731e
18 Sep, 2003 4 commits
- Add missing pthread_join function. This may have been responsible for · 531f48f2
  Moe Jette authored Sep 18, 2003
```
the left-over slurmctld process on abnormal termination.
```
  531f48f2
- Relocated pthread_join function to where the thread was created for greater · 158ccd3a
  Moe Jette authored Sep 18, 2003
```
clarity. No change in functionality or logic.
```
  158ccd3a
- Elaborate on the function of the "-c" option in the slurmctld man page. · caf9707d
  Moe Jette authored Sep 17, 2003
```
If the "-c" option is not specified then only the jobs and some node
state information will be preserved. Specifically the state of DOWN,
DRAINED, or DRAINING nodes and the associated reason field for those nodes.
```
  caf9707d
- Elaborate on the function of the "-c" option in the slurmctld man page. · 91876ed7
  Moe Jette authored Sep 17, 2003
  
  91876ed7
17 Sep, 2003 7 commits
- Clean-up of some error message. No change in logic. · bef92689
  Moe Jette authored Sep 17, 2003
  
  bef92689
- Don't crash on orphaned node · 9b7773e1
  jwindley authored Sep 17, 2003
  
  9b7773e1
- Beautify some error messages. No changes in logic. · 337676f9
  Moe Jette authored Sep 17, 2003
  
  337676f9
- Describe meanings of special job priorities: 0=held, · c8d7168e
  Moe Jette authored Sep 17, 2003
```
1=required nodes DOWN/DRAINED.
```
  c8d7168e
- Describe use of job priority zero and one (hold and required configuration · a58b0e00
  Moe Jette authored Sep 17, 2003
```
unavailable states).
```
  a58b0e00
- Set the priority of a job that can't run due to unavailable nodes to 1 · 187a8954
  Moe Jette authored Sep 17, 2003
```
right away (before running scheduling function).
```
  187a8954
- Raise the priority of a pending job when a node it was waiting for is · 5571c0a6
  Moe Jette authored Sep 17, 2003
```
returned to service. The priority is changed from 1 to value which would
be set for the job if submitted at that time.
(gnats:279)
```
  5571c0a6