Custom Query (196 matches)

This caused my rails app to 500 randomly (whenever the load balancer routes to b-b) due to problems with the fcgi dispatcher. Installing the gems locally seemed to make it work. Should we be ensuring that the gem build is consistent across each of the scripts servers, or is this something the user has to figure out on his own?

(Imported from ​help.mit.edu #405341.)

jbarnold:

For some reason,

rpm -e execsys; rpm -ivh execsys.rpm

on o-f does not populate binfmt_misc whereas the same command does populate binfmt_misc on b-m. Running "service execsys-binfmt start" manually works.

Fortunately, the confusion resulting from this bug did not cause any http downtime due to LVS.

Also: When o-f was not serving http requests, ssh-ing to scripts.mit.edu failed completely instead of allowing logins to the second server. Perhaps I didn't wait long enough.

It'd be awesome if we could set the status and owner of a ticket in the reply email. Geoff is working on this, and has some scrips that do part of the job.