Upgradingdave's questions

I'm working on building a feature for a larger system that will be able to parse and react to dsn bounce messages received from all different types of smtp servers.

It be great to test the code I write by reading a bunch of txt files containing examples of the types of bounce messages that might be encountered out in the wild ;-)

Anyone have any suggestions on where to find a bunch of test dsn messages lying around? Or any thoughts on how to create some good test data?

UPDATE: I found a ton of good test data inside this project: https://github.com/rjbs/mail-deliverystatus-bounceparser/tree/master/t/corpus

I'd like to configure apache so that http://my-domain.com/myapp serves a Python webapp running in CherryPy on a backend server.

Here's what's in the vhost:

    RewriteRule ^/myapp/?(.*) http://backend-server:8000/$1 [P]
    ProxyPassReverse /myapp/ http://backend-server:8000/

When I trace the request/response, I see:

GET /myapp HTTP/1.1
Host: my-domain.com

And then:

HTTP/1.1 303 See Other
Date: Thu, 15 Sep 2011 21:46:35 GMT
Server: CherryPy/3.1.2
Content-Type: text/html;charset=utf-8
Location: http://my-domain.com/somwhere-else/

As you can see, the CherryPy webapp sends a 303 redirect to /somewhere-else/

Any ideas why the Apache ProxyPassReverse doesn't transform the Location to http://my-domain.com/myapp/somewhere-else?

I've mounted a shared windows directory (C:\foo\bar) on an ubuntu server (/mnt/shared/bar) like so:

//windows-server/bar /mnt/shared/bar smbfs credentials=/etc/samba-credentials 0 0

It works great, but when I cd into /mnt/shared/bar, Ubuntu shows that all the files are owned by root.

I'd like to restrict access to /mnt/shared/bar on the linux side. For example, I'd like to only allow users in the "baz" group to have access to the /mnt/shared/bar directory.

I'm new to Samba. Can someone with more experience let me know if this is possible?

I have multiple directories spread over the file systems of multiple Ubuntu 10.04 servers.

The requirements are to allow a handful of people access a few of these directories over ftp.

For example joe needs to ftp files to /apps/app1. alice needs ability to ftp to /apps/app2 and /mnt/apps/app4 (nfs share to another server)

I really want to keep vsftpd chroot so that these users can't see anything but home directory.

All these users are ldap users. Their home directories are automounted from a central server.

So, ideally, when joe ftp's in, he should see a single directory named "app1". When alice ftps in, she should see directories named "app2" and "app4". Eventually, I'll need to add more "virtual" dirs. For example, alice might need access to /apps/app5 in a few weeks. I'd like to be able to add it so that when she ftp's in, she sees "app2", "app4" and "app5".

It seems like it should be easy, but I've been running in circles on how to get this to work.

OPTION 1

I've tried using fstab and/or autofs to mount stuff using --bind under user home directories. This kinda works. It's annoying that when you use autofs, the directories are not visible via ftp unless someone has accessed them recently.

Why OPTION 1 fails:

Same reason as OPTION 3

OPTION 2

I've tried using soft links, but again, this doesn't seem to work. I created a nfs mounted drive (using autofs) at /mnt/apps/app1. Then I create a link like so ln -s /mnt/apps/app1 /home/ldap/joe/app1.

Why OPTION 2 fails:

When I log into ftp as joe, I can see "app1" dir, but if I try to open it, I get: 550 Failed to open file. Error: Critical file transfer error. Note that /home/ldap/joe is an auto mounted drive that is physcially on a different server under /apps/exports/home/joe. I've read some other places that once a user is chrooted, they can't follow symbolic links outside their home dir.

OPTION3

Configure all users so they have the same home directory, then mount shared dirs under common home dir. If you use autofs, the dir names doen't show up when. If I use fstab, it's very close to a workable solution.

Why OPTION3 fails: I'm able to mount local dirs fine. But when I try to mount --bind <some mapped drive> /apps/app1, then when I ftp and browse the directory, I see a 500 Failed to change directory.

Anyone have an elegant solution for this problem? Maybe there's another ftp server that supports "virtual directories"?

I have an Ubuntu server running slapd. I've configured the same box as an ldap client. Logging into Ubuntu via LDAP works great as long as I follow these steps:

• Create group named 'testgroup1' using ldapaddgroup
• Create user named 'testuser1' using ldapadduser and add to testgroup1
• Then, I can successfully login to the server as testuser1 using ssh testuser1@my-server. This works great.

So, here's the problem: There are many existing entries in LDAP. Even though these existing entries are in the exact same ou as the 'testuser1' entry, I can't authenticate using any of the existing entries?

The existing entries only have objectClass=inetOrgPerson. So I added objectClass=posixAccount. But, when I try to ssh using these entries (that weren't created using ldapadduser), I see this in the debug output from nslcd:

nslcd: [3c9869] DEBUG: ldap_simple_bind_s("cn=existinguser,ou=my-ou,dc=example,dc=com","*****") (uri="ldap://<ip address>")
nslcd: [3c9869] failed to bind to LDAP server ldap://<ip address>: Invalid credentials: Success

The only difference I see between an entry created by ldapadduser and the existing entries, is that one has a structural objectClass of "Account" and the other has a structural objectClass of "inetOrgPerson".

So, I was wondering if anyone knows if it's required to use "Account" Structural objectClass in order to login to Ubuntu? Or can anyone think of any other reason nslcd is giving the error message above?

I posted a similar question over on super user

I'm trying to configure Ubuntu to login to ldap. I've come across several related posts, but I'm still confused:

How does pam_ldap find ldap users? Do I have to manually create an entry in /etc/passwd for every user in ldap?

The reason I ask is because I seem to be able to authenticate against ldap as long as I have a username inside the /etc/passwd file that corresponds to uid in ldap.

For example, user1 has a local account (that I created with adduser). There is also a ldap user uid=user1,ou=people,dc=mydomain,dc=com. The local password for user1 is different than the password for the ldap with the same uid. I can login with both passwords. Also, if I comment out the pam_unix line in common-auth, it will only let me in using the password defined in ldap.

So, I'm pretty sure pam_ldap is working. But when I try to log in using a user defined in ldap (but who does NOT have an entry in /etc/passwd), auth.log shows "invalid user".

So, I must be missing a step? Is there a way to sync ldap users into /etc/passwd? Or perhaps a config option to have pam_ldap create a corresponding local account when an ldap user first attempts to login?

It looks like the most recent version of apache available via aptitude is 2.2.14 (as shown here). However, we have a requirement from the customer to upgrade to 2.2.17, which is the latest version available on the apache site.

Is there a way to upgrade to version 2.2.17 via aptitude? If not, what's the best way to upgrade? I'm comfortable building from source, but would much prefer having the package manager take care of it if possible.

Also, I read on some other posts that the security vulnerabilities that have been fixed between 2.2.14 and 2.2.17 may have already been patched by the Ubuntu OS? If that's true, is there a way I can verify this on my system? Or just have to take Ubuntu's word for it?

Thanks for any advice!

Specifically, I was wondering if there are any performance benefits to compile apache2 from source in Ubuntu as opposed to installing via apt-get?