I have fish installed and working property (v1.23.0) on a set of 7 linux hosts (RHEL5). On 6 of those hosts there is a cron job that runs every minute of every day. That cron job executes a command similar to:
/usr/bin/ssh srv1 'perl /home/jason/some-script.pl' Normally, the command will produce no output, so cron does not send me an email of the results, which is as expected. But from time to time (5% perhaps), I will get output to stderr that is mailed to me: "fish: Could not connect to universal variable server, already tried manual restart (or no command supplied). You will not be able to share variable values between fish sessions. Is fish properly installed?" This seems to be caused when srv2-srv6 attempt to establish their ssh connections at or nearly at the exact same time. When this happens, it looks like multiple simultaneous attempts to start fishd on srv1 results in this error message. Now according to the ssh documentation, because I am specifying a command to be executed, that no login shell will be spawned, but obviously instances of 'fish' are being executed as non-login shells to handle the commands. I've read through the fish/fishd documentation for a configuration option to disable the use of the fishd process, but have not found any. I don't really need universal variables supported on this machine, so would be happy to disable the feature in order to avoid these error messages. Another question is: What sort of locking mechanism does fishd use to prevent mutliple instances from being launched or from conflicting with one another in such a way that leads to this error message, and can it be strengthened? Regards, Jason -- Jason L. Buberel ja...@buberel.org ------------------------------------------------------------------------------ _______________________________________________ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
