On Sat, 15 Mar 2014, Richard Guy Briggs wrote: > On 14/03/15, Robert P. J. Day wrote: > > > > currently playing with fedora (pre-21) rawhide, went to start > > mariadb server: > > > > # systemctl start mariadb-server > > > > and got (from output of "journalctl"): > > > > mariadb-check-socket[3942]: /usr/libexec/mariadb-check-socket: line 8: ./mariadb-scripts-common: No such file or directory > > systemd[1]: mariadb.service: control process exited, code=exited status=1 > > systemd[1]: Failed to start MariaDB database server. > > > > so i checked the script /usr/libexec/mariadb-check-socket and, sure > > enough, line 8: > > > > source ./mariadb-scripts-common > > What was the working directory of that script? which script? i invoked the "systemctl" command from a totally different directory, so i'm simply assuming that the mariadb-check-socket script has a working directory that is *not* /usr/libexec when it's being invoked; hence, the failure. > > but i'm fairly sure the "source" command uses the PATH variable to > > find scripts to source, > > Not if it is prefixed by "./". That will anchor that script in the pwd > of the running script (which could change during its execution). quite so, bad choice of words on my part. > > and there's no guarantee the above would work, even when that > > "common" script to be sourced is in the same directory as the > > sourcing script, isn't that correct? > > So say that script is in the same directory as the calling script > and in the calling script, a "cd <mumble>" command is executed? > > > as a hacky test, i edited the sourcing script and changed that > > line to: > > > > source /usr/libexec/mariadb-scripts-common > > I would use instead a local variable in the script that records the > starting working directory of that script and then prepend that > variable to that script name. right ... which is what other scripts in /usr/libexec seem to be doing. > > the above strikes me as simply a bug ... am i missing something > > here that would allow the original sourcing script to work > > properly? i'm about ready to file a bug report. > > I would certainly say it is a bug one way or another. i already filed it: https://bugzilla.redhat.com/show_bug.cgi?id=1076823 so we'll see what response i get. rday p.s. i was puzzled as to how something so trivially wrong(?) could have lasted for any length of time, but checking the git repo for mariadb source shows that it's a fairly recent addition. still, seems like something a really simple test should have caught. -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================