The dev and ops team are doing some performance tuning on the feedly Cloud to address some of the slowdowns identified last week. The service will be slower as a result today (Saturday April 23rd) and tomorrow (Sunday April 24th).
We might need to take the service offline for 2 hours later today to upgrade and restart some of the data nodes.
Sorry for the temporary inconvenience and have a great week end!
Update: Tuesday April 26 Morning – We are still suffering from waves of short 5-minute slowdowns / micro-outages. The dev team has increased the log level on our Hadoop cluster and servers to determine what is the component of the architecture which is responsible for those hiccups. We are also working on doubling the size of our servers in case we hit some limit somewhere. We will report back as soon as we have finished crunching through the logs (we generate about 100GB of logs for every hour of operation).
Note: collecting the logs and testing various hypothesis will likely continue to generate some slowness this morning, Wednesday morning and Thursday morning. Sorry for the inconvenience. We look forward to having a speedy feedly back.
/Seb, Kireet, and Edwin