Misleading "Another instance of BOINC is running" with wrong ownership of boinc dir

[eMPee584]

Just resolved this by picking the boinc pid from watch -d -n 0.2 systemctl status boinc-client.service, tracing the process within the retval = wait_client_mutex(".", 10); 10 second timeout with strace -fe trace=file -p $(pidof boinc) and that gave me the real issue:

command: "strace -fe trace=file -p 19601" (strace is /usr/bin/strace)
@dir: /home/empee584 @time: Thu  5 Oct 09:27:53 CEST 2017

strace: Process 19601 attached
open("./lockfile", O_WRONLY|O_CREAT, 0664) = -1 EACCES (Permission denied)
open("./lockfile", O_WRONLY|O_CREAT, 0664) = -1 EACCES (Permission denied)
open("./lockfile", O_WRONLY|O_CREAT, 0664) = -1 EACCES (Permission denied)
open("./lockfile", O_WRONLY|O_CREAT, 0664) = -1 EACCES (Permission denied)
open("./lockfile", O_WRONLY|O_CREAT, 0664) = -1 EACCES (Permission denied)
open("./lockfile", O_WRONLY|O_CREAT, 0664) = -1 EACCES (Permission denied)
open("/etc/localtime", O_RDONLY|O_CLOEXEC) = 3
+++ exited with 108 +++

"strace -fe trace=file -p 19601" exited after 6 seconds
exit status 0 at Thu  5 Oct 09:27:59 CEST 2017

which was caused by the fact that my older version of ranger doesn't transfer file permissions when copying a dir (ranger/ranger#734).
So the presented error message is wrong and totally misleading. Obviously, this is a corner case, but still. 😄

[makeasnek]

A bounty has been started by the SCI to fix this issue on BountySource. It is currently at $25 USD and will continue to increase until the issue has been fixed. You can contribute to this bounty directly if this issue is important to you.