We'll respond shortly.
Has anybody seen memcache clients suddenly stop being able to talk to the memcached server and mark it as dead?
In fact many people have seen this. The Pivot with the question is using the memcached gem, while others noted that the memcache-client gem had special retry logic within it to deal with such a situation. In the latter case, some have still seen connections to memcache timeout despite the retry logic and despite the fact that memcache appears to be up and operational by all other checks.
That being said, nobody had any notion of what the root cause was – anybody out on the interwebs have insight on this?
Anybody seen random segmentation faults and bus errors happening in Ruby?
One of our projects is seeing intermittent crashing of the Ruby interpreter (MRI) while running the app. It seems to happen in a random place inside of Rails, but never the exact same place. We can’t find anything suspicious about the code where it is crashing.
It was noted that the project in question is using libxml (to support Nokogiri) and that various combinations of libxml and MRI are known to be incompatible; in fact a few previous Blabs posts exist on this subject. However, in this case the crashes do not happen deterministically, which is unlike the cases described previously.