July 2015

Cooling events at IC

On the 12th & 16th of July, there were cooling failures at IC which may have caused disruption.

Poldi disk array

The RAID controller in poldi was swapped out for a more modern HBA capable of driving the drives at SATA3 speeds (see ticket). However, it seems that there are some reliability problems (ticket) which still need to be fixed.

Re-seat cables in ramoth

Ramoth (replica & backup database server) had been unstable, and the leading theory was loose cables. The RAID controller and riser were re-seated, RAM checked and contacts cleaned. Uptime seems much improved since then.


OWG had a meeting. TODO: link to summary.

Backend machine replacements

The backend web / API machines were upgraded from G5 (5th generation) to G7 machines, which brought improved CPUs (E5450 to X5650) with the AES-NI instruction to more efficiently handle SSL. Also, the newer machines are less likely to fail due to wearing out.