oxygen, a system in eqiad with role logging::kafkatee::webrequest::ops, is the box we use to store sampled 1:1000 webrequest logs, to be used in outage investigations and general purpose analysis by the SRE team.
The box has a purchasing date of 2011-01-27, i.e. almost 7 years ago, and thus way past its service life. We should replace it with a newer server, with similar specs. Ideally, switching to SSDs to be able run greps in the rare occasion they are needed, would be nice.
(Mark/Chris were working on a more comprehensive list of old > 5y servers, and probably more from this batch will get a procurement refresh task soon. I was just working on oxygen today and noticed how old it is, >3x slower my 2-year old laptop…)
@Ottomata/@elukey could provide additional information about the box, as well as storage requirements -- now it stores < 100GB, but perhaps they have needs or plans I haven't considered? We could also consider placing this into a VM, or even merging it with some other related role, like e.g. the syslog servers -- although I doubt these have SSD and are very fast either? Let's discuss and decide here :)