Commits · 3f6b4fda7115105e4da2f5c9ca8bd708251b6980 · Manuel G. Marciani / ces_slurm_simulator

13 Jun, 2003 6 commits
- updated setproctitle() calls to not add "slurmd" at the beginning of the · 3f6b4fda
  Mark Grondona authored Jun 13, 2003
```
 string (this is not needed).
```
  3f6b4fda
- o add new thread state : DSH_JOB_HUNG for nodes where job kill request · 1669e688
  Mark Grondona authored Jun 13, 2003
```
   failed (presumably due to unkillable processes)
 o retry failed JOB_KILL rpcs
```
  1669e688
- o do not allow repeated revokes of credentials · b474ac4b
  Mark Grondona authored Jun 13, 2003
  
  b474ac4b
- o removed unused "len" variable · c3dddde7
  Mark Grondona authored Jun 13, 2003
  
  c3dddde7
- o add string.h header to avoid warnings for strncpy · ef469c36
  Mark Grondona authored Jun 13, 2003
  
  ef469c36
- o add new error number ESLURMD_JOB_KILL_FAILED for failure of the job kill · 1b1b2a7a
  Mark Grondona authored Jun 13, 2003
```
   rpc.
```
  1b1b2a7a
12 Jun, 2003 10 commits
- Add cross references between man pages for squeue, sinfo, and scontrol. · 198770db
  Moe Jette authored Jun 12, 2003
  
  198770db
- o take neterr_mutex sooner to avoid hang in slurmd startup · aa01e5fc
  Mark Grondona authored Jun 12, 2003
  
  aa01e5fc
- Update man page to reflect new output formats and for field specification options. · a177ff45
  Moe Jette authored Jun 12, 2003
  
  a177ff45
- Add new output field options: %M = TIME_USED, %D= #NODES · 68c94d28
  Moe Jette authored Jun 12, 2003
```
Default output changed: Remove PRIO, TIME_LIMIT, Add TIME_USED #NODES
Long output changed: Remove PRIO, START_TIME, END_TIME. Add TIME_USED, #NODES
Step output changed: Remove START_TIME, Add TIME_USED
Output line "No jobs found in system" removed.
```
  68c94d28
- Don't wait for response from slurmd on reconfig request (was producing an · 804f30ab
  Moe Jette authored Jun 12, 2003
```
error).
```
  804f30ab
- Don't transition node from DRAIN to DOWN state when it registers with · aa13afa0
  Moe Jette authored Jun 12, 2003
```
too low configuration.
```
  aa13afa0
- Properly handle transisition of nodes from DOWN to a DRAIN* state. · 6705e2dd
  Moe Jette authored Jun 12, 2003
```
Formerly set state to DRAINING instead of DRAINED.
```
  6705e2dd
- Combine DRAINED and DRAINING nodes on a single line as state DRAIN unles · c4a0a5ff
  Moe Jette authored Jun 11, 2003
```
the --long option is specified.
```
  c4a0a5ff
- Fix bug in how KillWait default value is processed in config file. · 775ce2dc
  Moe Jette authored Jun 11, 2003
  
  775ce2dc
- Delete a job's steps when its completion is noted via *retry* of job kill · 88ac4b20
  Moe Jette authored Jun 11, 2003
```
RPC in agent.
```
  88ac4b20
11 Jun, 2003 19 commits
- o updated NEWS and META for slurm 0.2.5 · dafbaddb
  Mark Grondona authored Jun 11, 2003
  
  dafbaddb
- o add tests for setproctitle() emulation · 6ac33ad5
  Mark Grondona authored Jun 11, 2003
  
  6ac33ad5
- o add include for interconnect.h · dd161f39
  Mark Grondona authored Jun 11, 2003
  
  dd161f39
- o print Elan job info as a debug message · 413dada3
  Mark Grondona authored Jun 11, 2003
  
  413dada3
- o added qsw_capability_string function for easier printing of Elan job · 4963b2aa
  Mark Grondona authored Jun 11, 2003
```
  information
```
  4963b2aa
- o use setproctitle() to attempt process title change in slurmd. · cf519f62
  Mark Grondona authored Jun 11, 2003
```
 o add calls to interconnect_node_init/fini() in slurmd.
 o move interconnect_postfini() above wait_for_io()
```
  cf519f62
- o use kill(-pgid, 0) to determine if job's session is still running. · c0500712
  Mark Grondona authored Jun 11, 2003
  
  c0500712
- o fixes for kill job rpc. Only delay kill_wait seconds, and stop · a6505719
  Mark Grondona authored Jun 11, 2003
```
   waiting for processes once slurmctld closes the connection.
```
  a6505719
- o add empty handler for SIGTERM and SIGINT so the session manager · a4225278
  Mark Grondona authored Jun 11, 2003
```
   is not inadvertently killed.
 o slight changed in slurmd "task %d exited ..." debug message
```
  a4225278
- o New slurmd interconnect functions interconnect_node_init(), · 27783b02
  Mark Grondona authored Jun 11, 2003
```
   interconnect_node_fini() to allow for interconnect specific node
   initializations.
 o Elan interconnect module now runs the network error resolver thread
   (requires /etc/elanhosts config file)
```
  27783b02
- o Need to wait kill_wait + 2 seconds for slurmd reply to kill job request · eaa33e31
  Mark Grondona authored Jun 11, 2003
```
   instead of just kill_wait seconds because slurmd sleeps for kill_wait
   seconds, so therefore slurmctld would never recv a reply.
```
  eaa33e31
- o slurm_send_rc_msg() now has return value · a7a25e39
  Mark Grondona authored Jun 11, 2003
  
  a7a25e39
- o set kill_wait to default of 30s instead of NO_VAL · 0215309f
  Mark Grondona authored Jun 11, 2003
  
  0215309f
- Grow a name buffer as needed. · 04bd8854
  Moe Jette authored Jun 11, 2003
  
  04bd8854
- o print more specific error message when task exits with nonzero status · 17b5a572
  Mark Grondona authored Jun 11, 2003
  
  17b5a572
- o fixed bug recv_rc_msg() that resulted in "rc" always being 0. · 82be6f6a
  Mark Grondona authored Jun 11, 2003
```
 o Do not terminate job if job is already terminating, just exit launch thd.
```
  82be6f6a
- Fix typo "insure that a a job" -> "insure that a job". · 5584b2cd
  Moe Jette authored Jun 11, 2003
  
  5584b2cd
- Always print full node list, expand strings and fields as needed. · 7ebe30c0
  Moe Jette authored Jun 11, 2003
  
  7ebe30c0
- o send SIGHUP to IO thread before attempting pthread_join() in case there · 6a8bba19
  Mark Grondona authored Jun 11, 2003
```
   has been no IO activity since job ended/failed/etc.
```
  6a8bba19
10 Jun, 2003 2 commits
- o fix memory leak in plugrack_read_dir - no closedir() after opendir() · d8d07a31
  Mark Grondona authored Jun 10, 2003
  
  d8d07a31
- o fix memory leak in _send_recv_rc_msg() · a39e7049
  Mark Grondona authored Jun 10, 2003
  
  a39e7049
09 Jun, 2003 3 commits
- Hold default sinfo output to 80 columns. · 8303abad
  Moe Jette authored Jun 09, 2003
  
  8303abad
- Expand the sinfo hostlist field as much as needed if the "-l" option is · 4f8eace0
  Moe Jette authored Jun 09, 2003
```
specified.
```
  4f8eace0
- o do not abort() on fatal() calls, just exit · 94384552
  Mark Grondona authored Jun 09, 2003
```
 o do not call fatal() for lsd_fatal_error(), just call error()
```
  94384552