PGConf.Russia 2018

In this talk, I will tell you the story of how a bunch of sysadmins got sick of having to resuscitate their petabyte-sized Elasticsearch cluster and decided to replace it with some tried technologies: PostgreSQL, Kafka, a bit of Redis, lots of glue, and the typical sysadmin stubbornness. The result is Bagger: the sysadmin answer to Big Data. A fast, fairly reliable, fault-tolerant store, used mostly for logging timestamped events for some amount of time. Bagger is named the Bagger series of bucket-wheel excavators, feats of German engineering and some of the largest land vehicles ever produced by man. Just like the excavators that dig through tons of material, our Bagger digs through tons data.

When one faces the issues with PostgreSQL, the main suspicion falls on vacuum. Experience of Data Egret team proves how many DBAs are attacking this rake. While there are tons of information, documentation and discussions on vacuum itself, the topic is still associated with a lot of myths, tales, horror stories and misconceptions. In my talk I will try to reveal the key points concerning the inner structure of vacuum, basic approaches to its adjustment and tuning, performance monitoring, and so on.

This report is about my PostgreSQL extension pg_profile. This extension all you need to create periodic statistics snapshots and to keep them. You can build a report on one or many serial snapshots. This report will contain statistics information about database workload in specified time period. It is very useful to start investigation on performance degragation or excessive resoure consumption in the past.

It's not a secret for anyone that statistics can not be transferred with a major upgrade. For small and not heavily loaded databases this is not a problem, you can quickly collect new statistics. But we have databases with a volume of about 5TB and a load of about 100k rps, for which it became a big problem: taking off without statistics, the replicas could not even replay WAL. In my report I'll tell you what tricks we went to upgrade these databases with requirements of 100% read only availability, about what mistakes were made, and about how these errors were painfully corrected. The result of these errors was the extension called "pg_dirty_hands", in which we will collect various hacks, which can be last resort to repair data corruption.