[db-wg] Upgrading servers to address memory issues, intermittent outage in mailupdates

Dear working group,

Following the outage we experienced on 26 November we determined that a an in-memory cache was the cause. We have disabled this cache, and allocated more memory to the whois application process. We also determined that the whois process and the back-end database need more memory.

For this reason we are fast-tracking the deployment of new hardware and extending the memory in the servers that we use.

Unfortunately the deployment of two new servers led to a failure to deliver some of the mails sent to auto-dbm at ripe.net, the Mailupdates service for Whois, from approximately 12:00 UTC on Tuesday 6th December to 16:30 UTC on Wednesday 7th December. Any failure resulted in a reply with the Subject "Mail delivery failed: returning message to sender".

Kind regards

Tim Bruijnzeels

Assistant Manager Software Engineering
RIPE NCC Database Group