One ingredient to surviving [and "succeeding"] in science is having a few individuals write supportive letters of reference when you apply for jobs, tenure, and/or promotion. Previously, I posted some thoughts about Who should I get to write a letter of reference?
Avoid making last minute requests to people. This may lead to hastily written letters, no letter, or just "recycling" of old letters. It is worth thinking about who you may need or want to write a letter for you in the next year or so. Then maintain and/or cultivate that relationship. In particular, that means making sure they know what you have been up to scientifically for the last few years. It is idealistic to think that your former advisor/supervisor has been reading all your latest papers, particularly if (hopefully) you have moved into different areas. Hence, occasional update emails, visits, and chats at conferences are a good investment.
I suspect, that one unfortunate consequence of the rise of metrics is that letters are less influential than they used to be, except at the best institutions. Nevertheless, they still play a role.