Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
Menu
Open sidebar
OpenDAS
nni
Commits
e1ae623f
Unverified
Commit
e1ae623f
authored
Mar 22, 2019
by
SparkSnail
Committed by
GitHub
Mar 22, 2019
Browse files
Merge pull request #147 from Microsoft/master
merge master
parents
f796c60b
63697ec5
Changes
41
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
18 additions
and
15 deletions
+18
-15
tools/nni_trial_tool/trial_keeper.py
tools/nni_trial_tool/trial_keeper.py
+18
-15
No files found.
tools/nni_trial_tool/trial_keeper.py
View file @
e1ae623f
...
@@ -44,10 +44,9 @@ def main_loop(args):
...
@@ -44,10 +44,9 @@ def main_loop(args):
stdout_file
=
open
(
STDOUT_FULL_PATH
,
'a+'
)
stdout_file
=
open
(
STDOUT_FULL_PATH
,
'a+'
)
stderr_file
=
open
(
STDERR_FULL_PATH
,
'a+'
)
stderr_file
=
open
(
STDERR_FULL_PATH
,
'a+'
)
trial_keeper_syslogger
=
RemoteLogger
(
args
.
nnimanager_ip
,
args
.
nnimanager_port
,
'trial_keeper'
,
StdOutputType
.
Stdout
,
args
.
log_collection
)
trial_keeper_syslogger
=
RemoteLogger
(
args
.
nnimanager_ip
,
args
.
nnimanager_port
,
'trial_keeper'
,
StdOutputType
.
Stdout
)
# redirect trial keeper's stdout and stderr to syslog
# redirect trial keeper's stdout and stderr to syslog
trial_syslogger_stdout
=
RemoteLogger
(
args
.
nnimanager_ip
,
args
.
nnimanager_port
,
'trial'
,
StdOutputType
.
Stdout
)
trial_syslogger_stdout
=
RemoteLogger
(
args
.
nnimanager_ip
,
args
.
nnimanager_port
,
'trial'
,
StdOutputType
.
Stdout
,
args
.
log_collection
)
sys
.
stdout
=
sys
.
stderr
=
trial_keeper_syslogger
sys
.
stdout
=
sys
.
stderr
=
trial_keeper_syslogger
# backward compatibility
# backward compatibility
hdfs_host
=
None
hdfs_host
=
None
...
@@ -109,22 +108,25 @@ def check_version(args):
...
@@ -109,22 +108,25 @@ def check_version(args):
trial_keeper_version
=
pkg_resources
.
get_distribution
(
'nni'
).
version
trial_keeper_version
=
pkg_resources
.
get_distribution
(
'nni'
).
version
except
pkg_resources
.
ResolutionError
as
err
:
except
pkg_resources
.
ResolutionError
as
err
:
#package nni does not exist, try nni-tool package
#package nni does not exist, try nni-tool package
nni_log
(
LogType
.
Warning
,
'Package nni does not exist!'
)
nni_log
(
LogType
.
Error
,
'Package nni does not exist!'
)
try
:
os
.
_exit
(
1
)
trial_keeper_version
=
pkg_resources
.
get_distribution
(
'nni-tool'
).
version
except
pkg_resources
.
ResolutionError
as
err
:
#package nni-tool does not exist
nni_log
(
LogType
.
Error
,
'Package nni-tool does not exist!'
)
os
.
_exit
(
1
)
if
not
args
.
version
:
if
not
args
.
version
:
# skip version check
# skip version check
nni_log
(
LogType
.
Warning
,
'Skipping version check!'
)
nni_log
(
LogType
.
Warning
,
'Skipping version check!'
)
elif
trial_keeper_version
!=
args
.
version
:
nni_log
(
LogType
.
Error
,
'Exit trial keeper, trial keeper version is {}, and trainingService version is {},
\
versions does not match, please check your code and image versions!'
.
format
(
trial_keeper_version
,
args
.
version
))
os
.
_exit
(
1
)
else
:
else
:
nni_log
(
LogType
.
Info
,
'NNI version is {}'
.
format
(
args
.
version
))
regular
=
re
.
compile
(
'v?(?P<version>[0-9](\.[0-9]){0,2}).*'
)
try
:
trial_keeper_version
=
regular
.
search
(
trial_keeper_version
).
group
(
'version'
)
nni_log
(
LogType
.
Info
,
'trial_keeper_version is {0}'
.
format
(
trial_keeper_version
))
training_service_version
=
regular
.
search
(
args
.
version
).
group
(
'version'
)
nni_log
(
LogType
.
Info
,
'training_service_version is {0}'
.
format
(
training_service_version
))
if
trial_keeper_version
!=
training_service_version
:
nni_log
(
LogType
.
Error
,
'Version does not match!'
)
os
.
_exit
(
1
)
else
:
nni_log
(
LogType
.
Info
,
'Version match!'
)
except
AttributeError
as
err
:
nni_log
(
LogType
.
Error
,
err
)
if
__name__
==
'__main__'
:
if
__name__
==
'__main__'
:
'''NNI Trial Keeper main function'''
'''NNI Trial Keeper main function'''
...
@@ -141,6 +143,7 @@ if __name__ == '__main__':
...
@@ -141,6 +143,7 @@ if __name__ == '__main__':
PARSER
.
add_argument
(
'--nni_hdfs_exp_dir'
,
type
=
str
,
help
=
'nni experiment directory in hdfs'
)
PARSER
.
add_argument
(
'--nni_hdfs_exp_dir'
,
type
=
str
,
help
=
'nni experiment directory in hdfs'
)
PARSER
.
add_argument
(
'--webhdfs_path'
,
type
=
str
,
help
=
'the webhdfs path used in webhdfs URL'
)
PARSER
.
add_argument
(
'--webhdfs_path'
,
type
=
str
,
help
=
'the webhdfs path used in webhdfs URL'
)
PARSER
.
add_argument
(
'--version'
,
type
=
str
,
help
=
'the nni version transmitted from trainingService'
)
PARSER
.
add_argument
(
'--version'
,
type
=
str
,
help
=
'the nni version transmitted from trainingService'
)
PARSER
.
add_argument
(
'--log_collection'
,
type
=
str
,
help
=
'set the way to collect log in trialkeeper'
)
args
,
unknown
=
PARSER
.
parse_known_args
()
args
,
unknown
=
PARSER
.
parse_known_args
()
if
args
.
trial_command
is
None
:
if
args
.
trial_command
is
None
:
exit
(
1
)
exit
(
1
)
...
...
Prev
1
2
3
Next
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment