Excerpt |
---|
A snapshot of the UPS's used to support servers, switches, and other equipment under ChemIT's management. Mostly within ChemIT's Baker 248 server room. |
See also
UPS inventory for CCB Clusters and non-cluster HPCs
...
- Adding a USB connection from the UPS to a system allows that system to execute a shutdown command, properly shutting the system down in a timely manner during a prolonged power outage. Otherwise the system will simply lose power, and that kind of forced shutdown can often cause software and hardware failures.
- Ideally, ChemIT will have the resources to expand our capabilities to enable shutting down UPS-protected systems during prolonged power outages running beyond the capacities of the UPS, beyond the systems with a direct USB connection to the UPS.
Cluster name | UPS for | UPS shutdown algorithm, if any | Tools used | Other notes | |
---|---|---|---|---|---|
NONE | n/a | n/a | |||
(Unknown) | n/a | n/a | Cluster managed by CAC, not ChemIT | ||
Done Spring'14 |
| ||||
Done Spring'14 |
| ||||
Done Spring'14 |
| ||||
NONE | Merged with Widom cluster | n/a | n/a | ||
Scheraga: Current, production Matrix | Done Fall'14 |
| |||
Scheraga: Forthcoming Matrix | Done Fall'14 | UPS supporting both Synology storage system and headnode. UPS USB-connected to Synology storage system. Synology thus sends a signal to headnode. Algorithms are:
| Synology's own s/w. On Linux systems, running "nut". | ||
Widom (w/ Loring) | Done April 2016 |
| Moved Widom HeadNode to Loring UPS
| ||
ChemIT (C4) | Done |
| Moved C4 to Loring UPS | ||
Baird: 1 rack-mounted computational computer | NONE | n/a | n/a | ||
Freed: Eldor | NONE | n/a | n/a | ||
Petersen: 2 rack-mounted computational computers | Yes, but needs to be deployed in true production; using Widom's UPS for now. | UPS supporting both system #50 and system #51. UPS is USB-connected to system #50, which itself does not send signal to system #51. Algorithms for System #50 is: Shutdown if only 10% battery power is left. (System #51 currently does not have a way to be shutdown properly if there is a prolonged power outage.) | Windows OS | ChemIT would like to: Establish sending a signal from system #50 to system #51 and have system #51 properly shut down in the event of a prolonged outage. | |
Scheraga: 4 GPU rack-mounted computational computers | NONE | n/a | n/a |
Power outage impact on systems with and without UPS
~5-10 minute outage from Sunday, 4/23/207, per Michael Hint's investigations
Group or server | UPS info (details in above table) | Impact of outage: Headnode or main server
| Impact of outage: Storage | Impact of outage: Compute nodes (expect "down") | Impact of outage: Other |
---|---|---|---|---|---|
Chemistry IT: SERV-05: HyperV production hosts: Stockroom QB, Stockroom WebApp, ChemIT file share, test WSUS. (Dell, rack)
| Worthless: Died within 2 minutes. (Was a hand-me-down) | FAILED | Plan: All but ChemIT file share going to AWS. | ||
Chemistry IT: SERV-05: HyperV backup. (RedBarn, rack) | Worthless: Died within 2 minutes. (Was a hand-me-down) | FAILED | |||
RESE-01: HyperV hosts to CRANE-19 (NFS) Crane Synology | Survived | Fine | Fine | ||
Scheraga Matrix headnode Scheraga Matrix Synology | Survived | Fine | Fine | (down) | |
Hoffmann | Survived | Fine | n/a | (down) | Router config reset, so failed |
Lancaster- Crane | Survived | Fine | n/a | (down) | |
Widom-Loring-Abruna | Survived | Fine | n/a | bw001 up, since part of twin head node (all the rest were down) | |
Baird compute server | No UPS | Down (MH restarted remotely via IPMI) | n/a | ||
Petersen | Survived | ||||
Freed's Eldor | ? |
What does it cost to UPS a research system?
...
Most we been done Spring 14, after the spate of power failures. See CCB's HPC page (first chart, in "UPS for headnode" column) for details
Cluster | Done | Not done | Notes | Loring |
| X | Unique: Need to do ASAP|
---|---|---|---|---|---|---|---|
Abruna |
| X | Unique: Need to do ASAP |
...
See CCB's HPC page (second chart, in "UPS" column) and CCB's non-HPC page (in "UPS" column) for details of the few that are already done.
...