I have 2 t/ drives and the first fills to 100% the second sits at 0% and everything slows down. How can I fix this?
will the next plot move to nvme2?
swar view below
===========================================================================================================================
num job k plot_id pid start elapsed_time phase phase_times progress temp_size
===========================================================================================================================
1 samsung 32 3438142 17976 2021-05-24 15:49:44 06:30:52 3 02:13 / 01:34 54.73% 219 GiB
2 samsung 32 dc81a09 22891 2021-05-24 18:02:25 04:18:11 2 02:31 54.22% 179 GiB
3 samsung 32 7dac11a 24554 2021-05-24 19:02:30 03:18:06 1 10.09% 110 GiB
===========================================================================================================================
Manager Status: Running
=======================================================================
type drive used total % # temp dest
=======================================================================
t/- /mnt/nvme1 0.85TiB 0.89TiB 100.0% 3/- 1/2/3
t/- /mnt/nvme2 0.00TiB 0.89TiB 0.0% -/-
-/d /mnt/plot1 0.32TiB 7.22TiB 4.7% -/3 1/2/3
=======================================================================
CPU Usage: 7.1%
RAM Usage: 8.03/15.41GiB(55.0%)
Plots Completed Yesterday: 0
Plots Completed Today: 2
It seems you schedule all on nvme1 and nothing on nvme2… the % is the storage occupation (100% means it is full). From what I can see, you are full on nvme1, so the jobs are/will hang.
You can see on line:
t/- /mnt/nvme1 0.85TiB 0.89TiB 100.0% 3/- 1/2/3
This line shows that job 1/2/3 are running on nvme1 as temporary disk.
You should share your config file which is not good. Normally you should declare 2 job task on swar config : one for nvme1 and one for nvme2. each job will have its own stagger and plot queue limit.
Thanks for the reply, I’m not sure what I changed but it seems to be doing ok now. I have to do some more reading as I’m still not sure what I’m doing. I think with my H/W I should be getting more than 21 plots, just not sure what I need to change in my Swar config to optimise that?
===========================================================================================================================
num job k plot_id pid start elapsed_time phase phase_times progress temp_size
===================================================================================**======================================
1 samsung 32 2021-05-27 02:30:22 06:12:38 3 02:16 / 01:30 95.76% 159 GiB
2 samsung 32 2021-05-27 03:41:28 05:01:32 3 02:17 / 01:23 78.79% 160 GiB
3 samsung 32 2021-05-27 04:47:33 03:55:26 3 02:15 / 01:17 60.33% 210 GiB
4 samsung 32 2021-05-27 07:03:31 01:39:28 1 26.19% 167 GiB
5 samsung 32 2021-05-27 08:03:36 00:39:24 1 12.51% 142 GiB
===========================================================================================================================
Manager Status: Running
===========================================================================
type drive used total % # temp dest
===========================================================================
t/- /mnt/nvme1 0.52TiB 0.89TiB 61.7% 3/- 1/3/4
t/- /mnt/nvme2 0.43TiB 0.89TiB 50.8% 2/- 2/5
-/d /mnt/plot_1 3.98TiB 7.22TiB 58.1% -/5 1/2/3/4/5
===========================================================================
CPU Usage: 44.1%
RAM Usage: 14.02/15.41GiB(93.3%)
Plots Completed Yesterday: 21
Plots Completed Today: 7
`**indent preformatted text by 4 spaces**`