Central respository/database for Erlang Alarm reports
Currently, the acsys/erlang system sends emails to the person named in the config file. I literally have tens of thousands of these emails accumulated over the last few years. I think it would be better if there were a central repository for these notifications. I like to keep them around for historical purposes, so if someone notices something has been wrong with one node for a few days I can look back and look for anything interesting. With them collected in one place, it would also be easier to examine them and look at what is an outlier vs. the common stuff that happens every day (e.g. taking 54ms to reply to a read on $02).
#1 Updated by Richard Neswold 11 months ago
We should probably look at the new
logger module in Erlang that replaces
error_logger. You can add meta-data to the log messages and define several handlers which filter messages based on the meta-data. For instance, you could have debug-like messages go to the local drive for your own diagnostics but have urgent messages to go a notification system.
It would require some tweaking of our code to convert the old alarm handler stuff to the new stuff but it would be mostly mechanical.