which disk does compute node output get stored on?
49 views
Skip to first unread message
TJ
Jan 21, 2020, 11:35:11 PM1/21/20
to google-cloud-slurm-discuss
Hello everyone- I was wondering which disk the /home/[username] directory writes to. It seems that the output from autoscaled nodes gets written to some permanent boot disk, and I was wondering if it was the boot disk of the login node or the controller node. Thanks!
Wyatt Gorman
Jan 21, 2020, 11:42:20 PM1/21/20
to TJ, google-cloud-slurm-discuss
Hi TJ, this is stored on a NFS mount (run "mount" to see more info) running on the controller node. It is shared with the controller boot disk currently.
Wyatt Gorman HPC Solutions Manager |
On Tue, Jan 21, 2020 at 11:35 PM TJ <tjvon...@gmail.com> wrote:
Hello everyone- I was wondering which disk the /home/[username] directory writes to. It seems that the output from autoscaled nodes gets written to some permanent boot disk, and I was wondering if it was the boot disk of the login node or the controller node. Thanks!
--
You received this message because you are subscribed to the Google Groups "google-cloud-slurm-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-cloud-slurm-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-cloud-slurm-discuss/54407fc5-23eb-4cf2-ba24-c4a6eeb65d99%40googlegroups.com.
TJ
Jan 21, 2020, 11:46:58 PM1/21/20
to google-cloud-slurm-discuss
Thank you! Does this mean my login node boot disk can be small, like 20 gb, while I should make my controller node boot disk large?
On Tuesday, January 21, 2020 at 11:42:20 PM UTC-5, Wyatt Gorman wrote:
Hi TJ, this is stored on a NFS mount (run "mount" to see more info) running on the controller node. It is shared with the controller boot disk currently.
Wyatt Gorman
HPC Solutions Manager
On Tue, Jan 21, 2020 at 11:35 PM TJ <tjvon...@gmail.com> wrote:
Hello everyone- I was wondering which disk the /home/[username] directory writes to. It seems that the output from autoscaled nodes gets written to some permanent boot disk, and I was wondering if it was the boot disk of the login node or the controller node. Thanks!--
You received this message because you are subscribed to the Google Groups "google-cloud-slurm-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-cloud-slurm-discuss+unsub...@googlegroups.com.
Wyatt Gorman
Jan 21, 2020, 11:54:29 PM1/21/20
to TJ, google-cloud-slurm-discuss
Exactly right. The performance characteristics and possible tunings of the controller's Persistent disk that back that NFS server is specified in this document. Alternatively you could configure another external mount for your /home partition, or add a partition, using something like a Cloud Filestore, DDN's Lustre Marketplace offering, or an open source Lustre cluster so that you wouldn't rely on the controller for NFS. We'll have an update coming to Slurm-GCP shortly which will simplify this external storage mount configuration to a set of YAML fields as well.
On Tue, Jan 21, 2020 at 11:47 PM TJ <tjvon...@gmail.com> wrote:
Thank you! Does this mean my login node boot disk can be small, like 20 gb, while I should make my controller node boot disk large?
On Tuesday, January 21, 2020 at 11:42:20 PM UTC-5, Wyatt Gorman wrote:
Hi TJ, this is stored on a NFS mount (run "mount" to see more info) running on the controller node. It is shared with the controller boot disk currently.
Wyatt Gorman
HPC Solutions Manager
On Tue, Jan 21, 2020 at 11:35 PM TJ <tjvon...@gmail.com> wrote:
Hello everyone- I was wondering which disk the /home/[username] directory writes to. It seems that the output from autoscaled nodes gets written to some permanent boot disk, and I was wondering if it was the boot disk of the login node or the controller node. Thanks!--
You received this message because you are subscribed to the Google Groups "google-cloud-slurm-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-cloud-slurm-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-cloud-slurm-discuss/54407fc5-23eb-4cf2-ba24-c4a6eeb65d99%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "google-cloud-slurm-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-cloud-slurm-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-cloud-slurm-discuss/93891c53-9445-41e5-8e0b-5df897b26b7a%40googlegroups.com.
TJ
Jan 21, 2020, 11:55:13 PM1/21/20
to google-cloud-slurm-discuss
Thank you very much!
On Tuesday, January 21, 2020 at 11:54:29 PM UTC-5, Wyatt Gorman wrote:
Exactly right. The performance characteristics and possible tunings of the controller's Persistent disk that back that NFS server is specified in this document. Alternatively you could configure another external mount for your /home partition, or add a partition, using something like a Cloud Filestore, DDN's Lustre Marketplace offering, or an open source Lustre cluster so that you wouldn't rely on the controller for NFS. We'll have an update coming to Slurm-GCP shortly which will simplify this external storage mount configuration to a set of YAML fields as well.
Wyatt Gorman
HPC Solutions Manager
On Tue, Jan 21, 2020 at 11:47 PM TJ <tjvon...@gmail.com> wrote:
Thank you! Does this mean my login node boot disk can be small, like 20 gb, while I should make my controller node boot disk large?
On Tuesday, January 21, 2020 at 11:42:20 PM UTC-5, Wyatt Gorman wrote:
Hi TJ, this is stored on a NFS mount (run "mount" to see more info) running on the controller node. It is shared with the controller boot disk currently.
Wyatt Gorman
HPC Solutions Manager
On Tue, Jan 21, 2020 at 11:35 PM TJ <tjvon...@gmail.com> wrote:
Hello everyone- I was wondering which disk the /home/[username] directory writes to. It seems that the output from autoscaled nodes gets written to some permanent boot disk, and I was wondering if it was the boot disk of the login node or the controller node. Thanks!--
You received this message because you are subscribed to the Google Groups "google-cloud-slurm-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-cloud-slurm-discuss+unsub...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-cloud-slurm-discuss/54407fc5-23eb-4cf2-ba24-c4a6eeb65d99%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "google-cloud-slurm-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-cloud-slurm-discuss+unsub...@googlegroups.com.
Joseph Schoonover
Jan 22, 2020, 12:44:45 AM1/22/20
to google-cloud-slurm-discuss
In case you are interested, this marketplace offering (
https://console.cloud.google.com/marketplace/details/fluid-cluster-ops/fluid-slurm-gcp ) currently offers the ability to add/remove lustre and nfs mounts on-the-fly with an image based slurm-gcp.
https://console.cloud.google.com/marketplace/details/fluid-cluster-ops/fluid-slurm-gcp ) currently offers the ability to add/remove lustre and nfs mounts on-the-fly with an image based slurm-gcp.
Help documentation can be found at https://help.fluidnumerics.com/slurm-gcp
Reply all
Reply to author
Forward
0 new messages