Get rid of redundant caches
In some cases it is a very complicated and thus error-prone work to properly deal with caches. Moreover, if due to some programming error some cache is calculated incorrectly, one should recalculate this cache after a bug fix. Recalculating caches for production databases can take hours. So, we should remain caches just when they are strictly required, e.g. we can't compare error traces with patterns each time when we need to find these relations.
The issue was implemented in branch remove-caches.
Updated by Evgeny Novikov about 2 years ago
- Status changed from Resolved to Closed
This great improvement is in master since 29f75b3ec. As there is no more issues for Klever 1.1, that commit is marked as v1.1rc1. Let's everybody start extensive testing to reveal tricky bug and to issue Klever 1.1 in a week or less.
CI tests passed. Besides, I compare manually quality and speed of different actions before and after cache removing. I noticed that some bugs were fixed, e.g. at the moment after confirming mark associations corresponding numbers at the job page and at the jobs tree page are shown properly. Also, although in some cases I used different hardware for comparison, but after removing caches Bridge seems to operate even faster as before! Vladimir made very good optimizations.