Hello,
When looking up prices on the rate card available at this page, the rates for most instances correspond to the quote previewed when setting up, for example, a new JupyterLab. But in some cases there is a marked difference. For example, when requesting a mem2_ssd2_gpu1_x4, the rate card reports 0.5508 under the On-demand GBP/hr column while attempting to start a new instance, the cost per hr is quoted £5.5080.
[Image: Screenshot from 2023-01-03 14-33-34]
Can you please clarify this?
Thanks!
Thanks for reporting this. I did some testing today, and it seems that the system is charging at the rate appeared in our webUI rather than rate card which are 10 times difference for these new GPU instance. These instances are made available just last month, so this is the first time I got report on this. I suspect the rate in system is incorrect, but I need to confirm with project manager when they are back tomorrow. If it would be okay with you, could you try to avoid all mem2_ssd2_gpu1 until our investigation is finished? We do not want to over charge you, and I think we should get answer soon. I will follow up on it. The other instance types (including mem2_ssd1_gpu1) seems fine, so you can work on your pipeline development meanwhile.
0
Permanently deleted user
Thank you for looking into this!
0
Permanently deleted user
I think we already ran some overpriced jobs. Probably, not amounting to a lot but would it be possible to look into a refund procedure? Thanks!
I checked with product manager today. It turns out that the error is in file that the UI used for cost estimation before launch and real time cost calculation. However, the price in rate card and the billing system the same and correct. Therefore, you should be able to resume your work using these instance types. We will follow up on fixing the cost estimation in UI. Thanks again for reporting this.
0
Permanently deleted user
Thanks for following up! I found one job (Cloud Workstation) that I ran some time ago which used a mem2_ssd2_gpu1_x4 instance. The duration was 53 minutes and in the Monitor tab in the RAP, the Final Cost is reported as £4.8468. I think this looks consistent with what the supposedly wrong estimate was suggesting rather than with the rate card. Is this information also wrong or perhaps I am not reading this right? Just need to make sure before running longer jobs.
Yes, all the user-facing price for this type (mem2_ssd2) of instance are all wrong. They would all show up ~10x more expensive than it should be. However, the bill you receive would be correct and the same as rate card.
I think the confusion came from my first message. I saw real time cost correspond to estimate price, so I thought the billing would be wrong too. However, product manager told me they are separate.
0
Permanently deleted user
Hello, can you please confirm that also the amount that we see in the Profile- Billing section is wrong if I made use of those instances? I think that's also aligned with the "wrong" pricing which makes it a bit difficult to understand what the real bill will be. Is there any way we can see the correct bill before running big jobs?
Comments
9 comments
Thanks for reporting this. I did some testing today, and it seems that the system is charging at the rate appeared in our webUI rather than rate card which are 10 times difference for these new GPU instance. These instances are made available just last month, so this is the first time I got report on this. I suspect the rate in system is incorrect, but I need to confirm with project manager when they are back tomorrow. If it would be okay with you, could you try to avoid all mem2_ssd2_gpu1 until our investigation is finished? We do not want to over charge you, and I think we should get answer soon. I will follow up on it. The other instance types (including mem2_ssd1_gpu1) seems fine, so you can work on your pipeline development meanwhile.
Thank you for looking into this!
I think we already ran some overpriced jobs. Probably, not amounting to a lot but would it be possible to look into a refund procedure? Thanks!
I checked with product manager today. It turns out that the error is in file that the UI used for cost estimation before launch and real time cost calculation. However, the price in rate card and the billing system the same and correct. Therefore, you should be able to resume your work using these instance types. We will follow up on fixing the cost estimation in UI. Thanks again for reporting this.
Thanks for following up! I found one job (Cloud Workstation) that I ran some time ago which used a mem2_ssd2_gpu1_x4 instance. The duration was 53 minutes and in the Monitor tab in the RAP, the Final Cost is reported as £4.8468. I think this looks consistent with what the supposedly wrong estimate was suggesting rather than with the rate card. Is this information also wrong or perhaps I am not reading this right? Just need to make sure before running longer jobs.
Yes, all the user-facing price for this type (mem2_ssd2) of instance are all wrong. They would all show up ~10x more expensive than it should be. However, the bill you receive would be correct and the same as rate card.
I think the confusion came from my first message. I saw real time cost correspond to estimate price, so I thought the billing would be wrong too. However, product manager told me they are separate.
Hello, can you please confirm that also the amount that we see in the Profile- Billing section is wrong if I made use of those instances? I think that's also aligned with the "wrong" pricing which makes it a bit difficult to understand what the real bill will be. Is there any way we can see the correct bill before running big jobs?
Yes, that's the info I got. You can contact billing@dnanexus.com or ukbiobank-support@dnanexus.com to get help with real bill.
Please sign in to leave a comment.