I have read article https://tailscale.com/blog/case-of-spiky-file-descriptors/ where author investigates anomalies with the amount of open file descriptors.
He said that made snapshot of open files using lsof
which allowed him to track lifetime of file:
01:12:17.24 tailcontr 142810 ubuntu 273u IPv6 5725049 0t0 TCP 172.31.10.244:https->1.2.3.4:57666 (ESTABLISHED)
01:13:32.17 tailcontr 142810 ubuntu 273u IPv6 5725049 0t0 TCP 172.31.10.244:https->1.2.3.4:57666 (CLOSE_WAIT)
01:14:03.22 tailcontr 142810 ubuntu 273u sock 0,8 0t0 5725049 protocol: TCPv6
01:16:53.93 (socket is removed)
What flags should be used to track lifetime of file as given in the example above?
P.S: I want to investigate similar problem with TCPv6 connections. According to lsof -u <user>
I have a lot of files looking like:
01:14:03.22 tailcontr 142810 ubuntu 273u sock 0,8 0t0 5725049 protocol: TCPv6
And I want to understand where were these connections made to.