Server cooling issues (UCL)
We were notified on the 19th about an aircon failure at UCL, and had to shut down various machines to keep the temperature down:
- zark (service: jxapi.openstreetmap.org)
- draco (experimental: aerial imagery)
- ascalon (testing: ticket tracker)
- clifford (testing: forum)
All machines were powered up again by 26th.
Further cooling issues arose over the weekend of 29th-30th March starting with a dramatic rise in temperatures on the Friday evening which continued through Saturday. Three machines shut themselves down before we intervened and turned off all non-essential machines.
Correspondence with UCL on the Monday revealed that the aircon is in fact out of action (it sounds like it was never fixed after the first incident) and that it is unlikely to be fixed as they expect to be decanting the contents of that room to a new location as previously discussed.
Other things
- Upgraded CPUs in orm
- Ordered new machine to act as second Nominatim server
- Ordered two HP DL360 G6 machines to go to OUOSL ($4568). The initial use is for North America tile caches but can be repurposed later.