zclok010
|
7c94523eb2
|
backup
|
2019-05-23 11:46:49 +08:00 |
|
zclok010
|
6a7ef893c6
|
git ignore .vscode
|
2019-05-10 12:16:27 +08:00 |
|
zclok010
|
b130b4fd7f
|
Fix test case error introduced in commit b41e682 (excution filter change)
|
2019-05-09 15:15:48 +08:00 |
|
zclok010
|
d98eec6f15
|
Add CcpVersion and CustomProperties in register info
|
2019-05-07 10:43:33 +08:00 |
|
zclok010
|
0d44a14c3e
|
refine code
|
2019-05-06 20:04:10 +08:00 |
|
zclok010
|
d7a8e4a371
|
tidy log when cleaning up zombie tasks
|
2019-05-05 14:57:32 +08:00 |
|
zclok010
|
05138060e7
|
Fix a bug that processes in task are not actually terminated after task canceling if cgroup is not enable
|
2019-04-30 20:26:31 +08:00 |
|
zclok010
|
844bb394b9
|
Fix a bug that zombie task clean up would fail when nodemanager starts
|
2019-04-30 20:24:27 +08:00 |
|
zclok010
|
42b467d5af
|
tidy some code
|
2019-04-30 15:12:30 +08:00 |
FAREAST\chezhang
|
6483b16b0f
|
Update tasks' statistics before sending heartbeat to enable showing statistics when task is running
|
2019-04-26 10:19:23 +08:00 |
|
zclok010
|
b41e682ab3
|
Merge branch 'executionfilter' into v2
|
2019-04-16 15:42:04 +08:00 |
|
zclok010
|
2750349ba1
|
Add build-in execution filters to adjust task affinity in terms of core distribution in NUMA nodes and to modify command for preparation of mpi task
|
2019-04-16 15:29:26 +08:00 |
|
zclok010
|
8deeffa1f0
|
get home dir by tilde expansion
|
2019-04-08 15:26:17 +08:00 |
|
zclok010
|
2fa7452e49
|
fix a bug that memory is limited to the first NUMA node when using cgroup
|
2019-03-28 17:26:35 +08:00 |
|
zclok010
|
4e0e4a2d1e
|
change the default working directory to home
|
2019-03-28 17:02:54 +08:00 |
|
zclok010
|
7cfce24f5a
|
Change the owner of home directory, which is created by Linux nodemanager, to the user instead of leaving it root.
|
2019-03-26 13:42:02 +08:00 |
|
FAREAST\chezhang
|
6df8834638
|
Merge branch 'disableCgroup' into v2
|
2019-03-18 12:17:21 +08:00 |
|
chenling
|
1a477f059e
|
Fix a bug that HPC Pack environment variables will be lost if CCP_SWITCH_USER is set on CentOS
|
2019-03-14 17:02:11 +08:00 |
|
FAREAST\chezhang
|
ea1a7af86b
|
refine code
|
2019-03-13 17:23:23 +08:00 |
|
FAREAST\chezhang
|
bf49363154
|
Add environment variable CCP_DISABLE_CGROUP to enable running a task without cgroup
|
2019-03-13 17:18:38 +08:00 |
|
chenling
|
729f94e18e
|
fix Cores In Use not correct bug
|
2019-02-19 15:13:17 +08:00 |
|
chenling
|
c09de10d5e
|
fix Cores In Use not correct bug
|
2019-01-29 22:59:24 +08:00 |
|
FAREAST\chezhang
|
45eed2e48a
|
Update version info
|
2019-01-03 15:42:06 +08:00 |
|
FAREAST\chezhang
|
3604242371
|
Merge branch 'affinity' into v2
|
2019-01-03 15:37:17 +08:00 |
|
FAREAST\chezhang
|
0ba6267b9f
|
enable switching user and running task command by using environment variable CCP_SWITCH_USER, which is to fix a bug that command, like "ulimit -l" on Ubuntu, is executed as root even if the runas user is specified as other user
|
2019-01-03 15:36:31 +08:00 |
|
FAREAST\chezhang
|
14db4d87f3
|
Fix affinity bug
|
2019-01-02 21:11:51 +08:00 |
|
chenling
|
53211d728a
|
Update version info
|
2018-08-28 23:13:34 +08:00 |
|
FAREAST\chezhang
|
6b33a068d6
|
Reduce naming service request interval when failing to get response from headnode so that Linux compute nodes could appear in a shorter time after ARM template deployment finishes
|
2018-08-28 17:04:21 +08:00 |
|
FAREAST\chezhang
|
1055fec2ce
|
Fix a bug that nodemanager would delete user ssh keys
|
2018-07-31 16:22:59 +08:00 |
Evan Cui
|
2fea77493f
|
Upgrade the spdlog version; Build the nodemanager in old envi
|
2018-07-25 18:35:35 +08:00 |
|
FAREAST\chezhang
|
524161c851
|
Get Linux compute node Azure instance metadata - revise
|
2018-07-24 17:43:41 +08:00 |
|
FAREAST\chezhang
|
d9780148f7
|
Get Linux compute node Azure instance metadata
|
2018-07-17 21:13:41 +08:00 |
|
FAREAST\chezhang
|
710da06975
|
Revise a log message
|
2018-06-13 15:30:23 +08:00 |
Evan Cui
|
475627fd11
|
Fixed the version info
|
2018-05-16 17:58:37 +08:00 |
|
FAREAST\chezhang
|
bf9d5bf9ff
|
Generate hostfile/machinefile for various mpi applications
|
2018-04-20 20:27:32 +08:00 |
|
FAREAST\chezhang
|
d7023ec4d2
|
Fix a bug that task would fail when cgroup is not enabled
|
2018-03-12 21:13:29 +08:00 |
|
FAREAST\chezhang
|
77567a7b0d
|
Map Windows system account to Linux root user
|
2018-02-09 20:54:50 +08:00 |
|
Chenling Zhang
|
d90f79e8da
|
Retrieve public key from private key if its value is absent
|
2017-12-13 17:15:16 +08:00 |
|
Chenling Zhang
|
272193474d
|
#1690111 Avoid leaving credential information on disk when Linux node execution filter task fails - handle more exceptions
|
2017-11-03 18:07:09 +08:00 |
|
Chenling Zhang
|
f2e1b90aca
|
#1690111 Avoid leaving credential information on disk when Linux node execution filter task fails
|
2017-11-03 14:44:03 +08:00 |
|
Chenling Zhang
|
137e4ede1a
|
#1339398 do not monitor GPU info if initializing GPU driver fails.
|
2017-10-12 10:48:14 +08:00 |
|
Chenling Zhang
|
7ac024af9e
|
Fix a bug so that root user could pass mutual trust in mpi docker task.
|
2017-09-28 22:23:59 +08:00 |
|
evanc
|
3b9a07ce47
|
Fix the unit test failure, enabled cancellation token for classes, added version info
|
2017-09-26 08:30:19 -07:00 |
|
Chenling Zhang
|
1c5753b412
|
some revise
|
2017-09-26 17:22:24 +08:00 |
|
Chenling Zhang
|
3cfd2f0a43
|
Merge branch 'branch2'
|
2017-09-18 22:10:15 +08:00 |
|
Chenling Zhang
|
cf40cc19ae
|
some revise
|
2017-09-18 21:34:04 +08:00 |
|
Chenling Zhang
|
b56f10d590
|
Merge branch 'peektaskoutput'
|
2017-09-14 17:07:03 +08:00 |
|
Chenling Zhang
|
09dd23127b
|
some revise
|
2017-09-11 17:39:42 +08:00 |
|
Chenling Zhang
|
0b87ef5a11
|
fix code defect
|
2017-09-04 21:12:01 +08:00 |
|
Chenling Zhang
|
b3fcc87d75
|
peek task output for linux compute nodes
|
2017-09-04 21:06:34 +08:00 |