I’m happy to announce the release of a new piece of software that I’ve written. Cluster Tail or ctail for short. Basically, it multiplexes lots of SSH connections, to invoke tail -f on a large set of machines or log files.
It isn’t the most amazing idea, but if you have ever had hundreds of machines (or even a dozen), trying to understand what is going on across all of can be very difficult, and this is just another tool to make that easier.
Most other methods of reading log files on large clusters get away from classic Unix Philosophy, but ctail can easily be piped into other commands.
Have you looked at pdsh?
I don’t use it for tail’ing, but I often do things like “pdsh -a yum list updates” or “pdsh -a df -h” etc.
I know people running real clusters who say they couldn’t live without it.
Yeah, i use that and various parallel ssh/shell/pfor tools for doing general tasks, but for tail I wanted the ability to prefix the machine names to the output….
Now that’s an excellent idea. That’ll come in handy even with only 20 servers like we have
Thanks.
Thank you very much
I have a set of scripts named dssh,dscp (distributed ssh, and scp).
Anyway…. you can invoke commands on clusters of remote boxes and/or copy files to them.
I’ve also been working no a script called dtop which can collect metrics from all the machines in your cluster and how you their stats.
For example you could see MySQL seconds behind master on all the boxes or number of ‘items’ processed on all the boxes.
I need to get it on the other side of the fence though.
This looks cool too so I’m going to have to play with it……