Skip to content
GitLab
Menu
Projects
Groups
Snippets
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
Menu
Open sidebar
csbase-dev
sgarest-daemon
Commits
e2b6bc0e
Commit
e2b6bc0e
authored
Jan 18, 2021
by
Renato Figueiro Maia
Browse files
Melhorias nos logs gerados pelo daemon do SGA.
[SOMA-6476]
parent
1018da8d
Pipeline
#71445
failed with stages
in 1 minute and 16 seconds
Changes
3
Pipelines
2
Hide whitespace changes
Inline
Side-by-side
sga/application.lua
View file @
e2b6bc0e
...
...
@@ -159,6 +159,7 @@ local function startapp(self)
self
.
logger
:
debug
(
msg
)
local
res
,
err
=
fn
()
if
not
res
or
math.floor
(
res
/
100
)
==
4
then
self
.
logger
:
error
(
"Failed "
..
msg
..
": "
..
tostring
(
err
))
res
,
err
=
self
.
client
:
register
(
sga_type
,
self
.
driver
:
get_nodes
(),
self
.
server
)
if
not
res
then
self
.
logger
:
error
(
"Failed registering: "
..
err
)
...
...
sga/joblist.lua
View file @
e2b6bc0e
...
...
@@ -90,8 +90,9 @@ function joblist:terminate()
end
end
function
joblist
.
new
(
config
)
function
joblist
.
new
(
config
,
logger
)
local
self
=
{
logger
=
logger
,
filename
=
config
.
runtime_data_dir
..
"/sga_jobs_persisted_data."
..
config
.
driver
,
terminated
=
false
,
gc_coro
=
false
,
...
...
@@ -119,6 +120,7 @@ function joblist.new(config)
local
now
=
os.time
()
for
jid
,
job
in
pairs
(
jobs
)
do
if
job
.
done
and
now
>
job
.
done
+
60
then
self
.
logger
:
debug
(
"Job "
..
jid
..
" discarded 1 minute after it completed."
)
jobs
[
jid
]
=
nil
self
:
save
()
end
...
...
sga/server.lua
View file @
e2b6bc0e
...
...
@@ -42,11 +42,13 @@ local Server = safer.readonly {
if
not
job
.
done
then
table.insert
(
persisted_jobs
.
lost
,
{
cmd_id
=
job
.
cmd_id
}
)
cleanup_job
(
self
,
job
)
self
.
logger
:
error
(
"Job "
..
job
.
jid
..
" lost due to completion check failure: "
..
tostring
(
err
))
end
elseif
not
is_done
then
table.insert
(
persisted_jobs
.
retrieved
,
{
cmd_id
=
job
.
cmd_id
,
actions
=
create_action_uris
(
self
,
job
.
jid
)
}
)
else
cleanup_job
(
self
,
job
)
self
.
logger
:
info
(
"Job "
..
job
.
jid
..
" completed"
)
end
end
end
...
...
@@ -63,7 +65,7 @@ function server.new(config, logger, client, driver)
local
self
=
{
rs
=
restserver
:
new
():
host
(
config
.
sgad_bind_addr
or
"0.0.0.0"
):
port
(
config
.
sgad_port
),
joblist
=
joblist
.
new
(
config
),
joblist
=
joblist
.
new
(
config
,
logger
),
base_uri
=
"http://"
..
config
.
sgad_host
..
":"
..
config
.
sgad_port
,
config
=
config
,
logger
=
logger
,
...
...
@@ -90,14 +92,15 @@ function server.new(config, logger, client, driver)
copas
.
sleep
(
self
.
config
.
register_retry_s
)
end
cleanup_job
(
self
,
job
)
self
.
logger
:
info
(
"
Command
"
..
job
.
jid
..
"
done
"
)
self
.
logger
:
info
(
"
Job
"
..
job
.
jid
..
"
completed
"
)
else
cleanup_job
(
self
,
job
)
self
.
logger
:
warn
(
"Job "
..
job
.
jid
..
" completed, but not reported due to lack of execution time information"
)
end
break
elseif
is_done
==
nil
then
self
.
logger
:
error
(
walltime_s
)
cleanup_job
(
self
,
job
)
self
.
logger
:
error
(
"Job "
..
job
.
jid
..
" lost due to completion check failure: "
..
tostring
(
walltime_s
))
break
end
end
...
...
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