Takipi/Overops is a must!
Overops (formerly Takipi) is a great tool that has been a fantastic asset to our debugging capabilities. Super easy to set up, and we've noticed no issues with the minimal overhead introduced. There are many ties where we've been able to solve a problem without having to try to reproduce it just because we had Overops installed.
Easy to use, very focused use case and extremely helpful information.
A bit pricey for a small company with a server per customer to monitor. Probably ok if you have one server for many customers.
Brilliant concept, updates occasionally crash JVM, price gougers
The customer support is generally very good. The concept is outstanding. Being able to record runtime state in production for post-mortem debugging is utterly amazing.
There are a number of critical application issues described below, but the chief con of Overops is the price. This is *at least* an order of magnitude more than the value that Overops provides, and is a kick in the teeth for surely most startup-level organisations, particularly ones like ours that helped them grow their business.
There were at least 3 occasions in 2+ years of use that Overops (previously Takipi) released updates to their agent (which runs in our production JVM) that caused the whole jvm to segfault. This is a pretty poor track record (and it's more than a little ironic that this happens with a *software quality tool*).
In all that time we never succeeded in getting a noise-free or complete exception notification channel between Overops and our Slack installation. It would fail to notify about most errors, and routinely report errors we had attempted to filter out. This persisted despite numerous sessions with app engineers.