Skip to content
GitLab
Explore
Sign in
Primary navigation
Search or go to…
Project
S
scord
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Snippets
Build
Pipelines
Jobs
Pipeline schedules
Artifacts
Deploy
Releases
Package Registry
Model registry
Operate
Environments
Terraform modules
Monitor
Incidents
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
eu
ADMIRE
scord
Commits
9ed93496
Commit
9ed93496
authored
11 months ago
by
Ramon Nou
Browse files
Options
Downloads
Patches
Plain Diff
Solving root installation defines
parent
6152a5f2
No related branches found
No related tags found
1 merge request
!126
Resolve "Root issues and stage-out with SPANK"
Pipeline
#4451
failed
11 months ago
Changes
3
Pipelines
1
Hide whitespace changes
Inline
Side-by-side
Showing
3 changed files
plugins/adhoc_services.d/gekkofs.sh
+16
-8
16 additions, 8 deletions
plugins/adhoc_services.d/gekkofs.sh
plugins/slurm/scord_prolog.sh.in
+15
-4
15 additions, 4 deletions
plugins/slurm/scord_prolog.sh.in
plugins/slurm/slurmadmcli.c
+4
-4
4 additions, 4 deletions
plugins/slurm/slurmadmcli.c
with
35 additions
and
16 deletions
plugins/adhoc_services.d/gekkofs.sh
+
16
−
8
View file @
9ed93496
#!/usr/bin/bash
echo
"GEKKOFS Script Called"
$HOSTNAME
$SLURM_JOBID
# If GKFS_DAEMON is not defined then define it here
if
[
-z
"
$GKFS_DAEMON
"
]
;
then
GKFS_DAEMON
=
/home/rnou/iodeps/bin/gkfs_daemon
fi
# If LIBGKFS_HOSTS_FILE is not defined then define it here
if
[
-z
"
$LIBGKFS_HOSTS_FILE
"
]
;
then
LIBGKFS_HOSTS_FILE
=
/tmp/gekkofs/gkfs_hosts.txt
fi
if
[
"
$1
"
==
"start"
]
;
then
echo
"Starting GEKKOFS"
nodes
=
$3
num_nodes
=
$(
echo
$nodes
|
awk
-F
,
'{print NF}'
)
# If num_nodes is
greater than 4
0, we are on the testing environment
if
[
$num_nodes
-
gt
4
0
]
;
then
# If num_nodes is
5
0, we are on the testing environment
if
[
$num_nodes
-
eq
5
0
]
;
then
exit
0
fi
workdir
=
$5
datadir
=
$7
mountdir
=
$9
unset
SLURM_CPU_BIND SLURM_CPU_BIND_LIST SLURM_CPU_BIND_TYPE SLURM_CPU_BIND_VERBOSE
srun
-N
$num_nodes
-n
$num_nodes
--oversubscribe
--overlap
--cpus-per-task
=
1
--mem-per-cpu
=
1
--export
=
ALL bash
-c
"mkdir -p
$mountdir
; mkdir -p
$datadir
"
srun
-N
$num_nodes
-n
$num_nodes
--oversubscribe
--overlap
--cpus-per-task
=
4
--mem-per-cpu
=
1
--export
=
ALL bash
-c
"gkfs_daemon --rootdir
$datadir
--mountdir
$mountdir
"
&
srun
-N
$num_nodes
-n
$num_nodes
--oversubscribe
--overlap
--cpus-per-task
=
1
--mem-per-cpu
=
1
--export
=
ALL /usr/bin/bash
-c
"mkdir -p
$mountdir
; mkdir -p
$datadir
"
srun
-N
$num_nodes
-n
$num_nodes
--oversubscribe
--overlap
--cpus-per-task
=
1
--mem-per-cpu
=
1
--export
=
ALL /usr/bin/bash
-c
"
$GKFS_DAEMON
--rootdir
$datadir
--mountdir
$mountdir
-H
$LIBGKFS_HOSTS_FILE
"
&
sleep
4
elif
[
"
$1
"
==
"stop"
]
;
then
echo
"Stopping GEKKOFS"
nodes
=
$3
num_nodes
=
$(
echo
$nodes
|
awk
-F
,
'{print NF}'
)
# If num_nodes is
greater than 4
0, we are on the testing environment
if
[
$num_nodes
-
gt
4
0
]
;
then
# If num_nodes is
5
0, we are on the testing environment
if
[
$num_nodes
-
eq
5
0
]
;
then
exit
0
fi
unset
SLURM_CPU_BIND SLURM_CPU_BIND_LIST SLURM_CPU_BIND_TYPE SLURM_CPU_BIND_VERBOSE
srun
-N
$num_nodes
-n
$num_nodes
--overlap
--oversubscribe
--cpus-per-task
=
1
--mem-per-cpu
=
1
--export
=
ALL bash
-c
"pkill -9 gkfs_daemon"
srun
-N
$num_nodes
-n
$num_nodes
--overlap
--oversubscribe
--cpus-per-task
=
1
--mem-per-cpu
=
1
--export
=
ALL
/usr/bin/
bash
-c
"pkill -9 gkfs_daemon"
elif
[
"
$1
"
==
"expand"
]
;
then
echo
"Expand command"
elif
[
"
$1
"
==
"shrink"
]
;
then
...
...
This diff is collapsed.
Click to expand it.
plugins/slurm/scord_prolog.sh.in
+
15
−
4
View file @
9ed93496
...
...
@@ -176,11 +176,21 @@ CARGO_ID=$(echo "cargo_$SLURM_JOB_ID.$SLURM_JOB_UID" | sha256sum | awk '{ print
CARGO_CONFIG_FILE
=
$CARGO_CONFIG_DIRECTORY
/
$CARGO_ID
.cfg
CARGO_MASTER_ADDRESS
=
"
$SCORDCTL_PROTO
://
$ADDRESS
:
$CARGO_PORT
"
CARGO_INSTANCE_NAME
=
$(
systemd-escape
--template
cargo@.service
"
$CARGO_ID
"
)
# This will fail always as we do not have the job registered in this moment
#if ! CARGO_NUM_NODES=$(@SCORD_QUERY_PROGRAM@ -s @SCORD_SERVICE_ADDRESS@ "$SLURM_JOB_ID" | grep io_procs | awk '{ print $2 }'); then
# echo "Failed to determine the number of I/O processes for job $SLURM_JOB_ID"
#else
CARGO_NUM_NODES
=
${#
hostnames
[@]
}
#fi
# If LIBGKFS_HOSTS_FILE is nor defined then do it
if
[
-z
"
$LIBGKFS_HOSTS_FILE
"
]
;
then
LIBGKFS_HOSTS_FILE
=
/tmp/gekkofs/gkfs_hosts.txt
fi
if
!
CARGO_NUM_NODES
=
$(
@SCORD_QUERY_PROGRAM@
-s
@SCORD_SERVICE_ADDRESS@
"
$SLURM_JOB_ID
"
|
grep
io_procs |
awk
'{ print $2 }'
)
;
then
echo
"Failed to determine the number of I/O processes for job
$SLURM_JOB_ID
"
else
CARGO_NUM_NODES
=
${#
hostnames
[@]
}
# if number of CARGO_NUM_NODES is below 2, use 2, they will be colocated
if
[
$CARGO_NUM_NODES
-lt
2
]
;
then
CARGO_NUM_NODES
=
2
fi
cat
<<
EOT
>>"
$CARGO_CONFIG_FILE
"
...
...
@@ -188,6 +198,7 @@ CARGO_ID=$CARGO_ID
CARGO_HOSTS=
$hostnames_csv
CARGO_NUM_NODES=
$CARGO_NUM_NODES
CARGO_ADDRESS=
$CARGO_MASTER_ADDRESS
LIBGKFS_HOSTS_FILE=
$LIBGKFS_HOSTS_FILE
EOT
CUID
=
$(
id
-u
$SLURM_JOB_USER
)
chown
"
$SLURM_JOB_USER
"
:
"
$SLURM_JOB_GROUP
"
"
$CARGO_CONFIG_FILE
"
...
...
This diff is collapsed.
Click to expand it.
plugins/slurm/slurmadmcli.c
+
4
−
4
View file @
9ed93496
...
...
@@ -953,11 +953,11 @@ scord_unregister_job(spank_t sp, scord_plugin_config_t cfg,
}
// remove_adhoc_storage
ADM_remove_adhoc_storage
(
scord_server
,
adhoc_storage
);
// remove all the files (this should be done on all the nodes.. TODO)
ADM_terminate_adhoc_storage
(
scord_server
,
adhoc_storage
);
//
ADM_remove_adhoc_storage(scord_server, adhoc_storage);
//
remove all the files (this should be done on all the nodes.. TODO)
remove_dir_content
(
adhoc_path
);
rmdir
(
adhoc_path
);
// remove job
ADM_remove_job
(
scord_server
,
scord_job
);
...
...
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment