Custom Query (196 matches)

(Imported from ​help.mit.edu #547795.)

abhagi:

Also, can one upload JSP files onto scripts.mit.edu?

geofft:

We do not run a Tomcat installation the same way we run an Apache httpd installation on scripts; Tomcat does not allow us to do security isolation between users in the same manner, so we cannot set up a multiuser installation. There is an Apache-httpd module (mod_proxy_ajp) that appears to allow running JSP scripts, but we are not yet sure whether this properly and securely isolates user accounts. We will try to take a look at this sometime soon and determine whether this method is a practical solution; other users have expressed some interest in running JSP pages on scripts.

Most of our software supports upgrading by switching to a newer tag, updating, and possibly running an upgrade databases script. If we install a SVN (or whatever VCS) working copy rather than extracting a tarball, users can upgrade their own installs easily. Our auto-upgrade process should also become easier.

For SVN, we can merely generate a tarball with the .svn directories and keep our existing auto-install (and to some extent, auto-upgrade) infrastructure. I suspect this will work with all other VCSes.

There are also packages such as Django that very much recommend installing HEAD rather than a tarballed release, so having an infrastructure that will deal with VCSes will make it easier to support such packages.

When I run parallel-find.pl on my OS X, 10.4, laptop and it does a find in a sipb-cell locker, the find tends to hang. Right now I have about fifty finds in uninterruptible wait, hours later, despite kill -9'ing them.

One of them's even in the net cell.

This may have something to do with me holding price.root tokens rather than price; I don't normally have trouble using the sipb cell, of course.

Does this problem afflict those of you who've tried parallel-find.pl from other systems?