Mei's questions

In trying to install scsh (not ssh and not csh) - the Scheme Shell - on Ubuntu 15.10, I keep running into dependency errors and it won't let me do it:

emd@luna:~$ sudo apt-get install scsh
Reading package lists... Done
Building dependency tree       
Reading state information... Done
Some packages could not be installed. This may mean that you have
requested an impossible situation or if you are using the unstable
distribution that some required packages have not yet been created
or been moved out of Incoming.
The following information may help to resolve the situation:

The following packages have unmet dependencies:
 scsh : Depends: scsh-0.6 but it is not installable
E: Unable to correct problems, you have held broken packages.
emd@luna:~$ 

Looking at what my installed system knows about scsh, I get:

emd@luna:~$ sudo apt-cache search scsh
scsh - A `scheme' interpreter designed for writing system programs (default version)
scsh-0.6-doc - Documentation for scsh, "The Scheme Shell"
scsh-common-0.6 - A `scheme' interpreter designed for writing system programs
scsh-doc - Documentation for scsh, "The Scheme Shell"
scsh-install-lib - scsh package installer
scsh-0.6 - A `scheme' interpreter designed for writing system programs

And about scsh-0.6 I get:

emd@luna:~$ sudo apt-cache show scsh-0.6
Package: scsh-0.6
Priority: extra
Section: universe/interpreters
Installed-Size: 8148
Maintainer: Ubuntu Developers <[email protected]>
Original-Maintainer: Scheme48 Maintainers Task Force <[email protected]>
Architecture: i386
Version: 0.6.7-8
Replaces: scsh (<= 0.6.6-3)
Depends: libc6 (>= 2.11), scsh-common-0.6 (= 0.6.7-8)
Suggests: scsh-0.6-doc
Conflicts: scsh (<= 0.6.6-3)
Filename: pool/universe/s/scsh-0.6/scsh-0.6_0.6.7-8_i386.deb
Size: 2996714
MD5sum: d736fabde3780e1fbab357ff12999887
SHA1: 97824c4f80f90e722b2be2fc0c77335fad293ed2
SHA256: c73e54f75506a431cd1672534b1760495a690b8688aee779410cead6fd395789
Description-en: A `scheme' interpreter designed for writing system programs
 Scsh has a high-level process notation for doing shell-script like
 tasks: running programs, establishing pipelines and I/O redirection.
 Scsh embeds this process notation within a full implementation of
 Scheme, a minimal and clean dialect of the Lisp programming language.
 The process notation is realized as a set of macro definitions, and
 is carefully designed to allow full integration with standard Scheme
 code. Scsh isn't Scheme-like; it is Scheme.
 .
 At the scripting level, scsh also has an Awk design, also implemented
 as a macro that can be embedded inside general Scheme code.
 .
 Scsh additionally provides the low-level access to the operating
 system normally associated with C. The current release provides full
 access to POSIX, plus important non-POSIX extensions, such as
 complete sockets support. "Full access to POSIX" means: fork, exec &
 wait, sockets, full read, write, open & close, seek & tell, complete
 file-system access, including stat, chmod/chgrp/chown, symlink, FIFO
 & directory access, tty & pty support, file locking, pipes, select,
 file-name pattern-matching, time & date, environment variables,
 signal handlers, and more.
Description-md5: 8ff47bffe440b87548670b2020e9a4db
Homepage: http://www.scsh.net/
Bugs: https://bugs.launchpad.net/ubuntu/+filebug
Origin: Ubuntu

And looking at the policy for scsh I get:

emd@luna:~$ sudo apt-cache policy scsh-0.6
scsh-0.6:i386:
  Installed: (none)
  Candidate: 0.6.7-8
  Version table:
     0.6.7-8 0
        500 http://mirror.team-cymru.org/ubuntu/ wily/universe i386 Packages

Looking at the source mirror URL, I see the following valid URLs:

1. http://mirror.team-cymru.org/ubuntu/pool/universe/s/scsh-0.6/scsh-0.6_0.6.7-8_i386.deb
2. http://mirror.team-cymru.org/ubuntu/pool/universe/s/scsh-0.6/scsh-0.6-doc_0.6.7-8_all.deb
3. http://mirror.team-cymru.org/ubuntu/pool/universe/s/scsh-0.6/scsh-common-0.6_0.6.7-8_all.deb

Trying to install those as a block (manually after downloading) - I get this:

emd@luna:~/Downloads$ sudo dpkg -i scsh*
Selecting previously unselected package scsh-0.6:i386.
(Reading database ... 220195 files and directories currently installed.)
Preparing to unpack scsh-0.6_0.6.7-8_i386.deb ...
Unpacking scsh-0.6:i386 (0.6.7-8) ...
Selecting previously unselected package scsh-0.6-doc.
Preparing to unpack scsh-0.6-doc_0.6.7-8_all.deb ...
Unpacking scsh-0.6-doc (0.6.7-8) ...
Selecting previously unselected package scsh-common-0.6.
Preparing to unpack scsh-common-0.6_0.6.7-8_all.deb ...
Unpacking scsh-common-0.6 (0.6.7-8) ...
dpkg: dependency problems prevent configuration of scsh-0.6:i386:
 scsh-0.6:i386 depends on scsh-common-0.6 (= 0.6.7-8).

dpkg: error processing package scsh-0.6:i386 (--install):
 dependency problems - leaving unconfigured
Setting up scsh-0.6-doc (0.6.7-8) ...
Setting up scsh-common-0.6 (0.6.7-8) ...
Processing triggers for man-db (2.7.4-1) ...
Processing triggers for doc-base (0.10.6) ...
Processing 3 added doc-base files...
Registering documents with scrollkeeper...
Errors were encountered while processing:
 scsh-0.6:i386

My question is: How do I install scsh fully?

A secondary question is: How do I resolve this sort of problem when it comes up? What further investigation can I do?

Update: (more information)

Looking at scsh-common, I see this:

emd@luna:~/Downloads$ sudo apt-cache show scsh-common-0.6
Package: scsh-common-0.6
Priority: extra
Section: universe/interpreters
Installed-Size: 1908
Maintainer: Ubuntu Developers <[email protected]>
Original-Maintainer: Scheme48 Maintainers Task Force <[email protected]>
Architecture: all
Source: scsh-0.6
Version: 0.6.7-8
Replaces: scsh (<= 0.6.6-3)
Suggests: scsh-0.6-doc
Conflicts: scsh (<= 0.6.6-3)
Filename: pool/universe/s/scsh-0.6/scsh-common-0.6_0.6.7-8_all.deb
Size: 378248
MD5sum: e75d6b286a8490482bc410f32601e24e
SHA1: 3ee3fe160efe72d999f8dda1665163db725dc9f1
SHA256: ba74ac8d9dae76e988bdc60054b808f3a6e6a175c7d00724560187a2729870a8
Description-en: A `scheme' interpreter designed for writing system programs
 Scsh has a high-level process notation for doing shell-script like
 tasks: running programs, establishing pipelines and I/O redirection.
 Scsh embeds this process notation within a full implementation of
 Scheme, a minimal and clean dialect of the Lisp programming language.
 The process notation is realized as a set of macro definitions, and
 is carefully designed to allow full integration with standard Scheme
 code. Scsh isn't Scheme-like; it is Scheme.
 .
 At the scripting level, scsh also has an Awk design, also implemented
 as a macro that can be embedded inside general Scheme code.
 .
 Scsh additionally provides the low-level access to the operating
 system normally associated with C. The current release provides full
 access to POSIX, plus important non-POSIX extensions, such as
 complete sockets support. "Full access to POSIX" means: fork, exec &
 wait, sockets, full read, write, open & close, seek & tell, complete
 file-system access, including stat, chmod/chgrp/chown, symlink, FIFO
 & directory access, tty & pty support, file locking, pipes, select,
 file-name pattern-matching, time & date, environment variables,
 signal handlers, and more.
 .
 This package contains the architecture-independent files.
Description-md5: b67b8d361914d72dc0e5b30aaca2650b
Homepage: http://www.scsh.net/
Bugs: https://bugs.launchpad.net/ubuntu/+filebug
Origin: Ubuntu

The installation process is marked by dpkg this way:

emd@luna:~/Downloads$ dpkg -l | sed -n '1,5p; /scsh/p'
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name                                                        Version                                    Architecture Description
+++-===========================================================-==========================================-============-===============================================================================
iU  scsh-0.6:i386                                               0.6.7-8                                    i386         A `scheme' interpreter designed for writing system programs
ii  scsh-0.6-doc                                                0.6.7-8                                    all          Documentation for scsh, "The Scheme Shell"
ii  scsh-common-0.6                                             0.6.7-8                                    all          A `scheme' interpreter designed for writing system programs

Only differences I see is that the package scsh wants scsh-common-0.6:i386 but the provided package is scsh-common-0.6:all... but that shouldn't matter should it?

Also, I'm also running:

emd@luna:~/Downloads$ cat /etc/*release*
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=15.10
DISTRIB_CODENAME=wily
DISTRIB_DESCRIPTION="Ubuntu 15.10"
NAME="Ubuntu"
VERSION="15.10 (Wily Werewolf)"
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME="Ubuntu 15.10"
VERSION_ID="15.10"
HOME_URL="http://www.ubuntu.com/"
SUPPORT_URL="http://help.ubuntu.com/"
BUG_REPORT_URL="http://bugs.launchpad.net/ubuntu/"
emd@luna:~/Downloads$ uname -a
Linux luna 4.2.0-23-generic #28-Ubuntu SMP Sun Dec 27 17:47:31 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux

I have a Samba machine - set up as a standalone server, and it's not showing when browsing the network. The workgroup is OFFICE (yah, I know.... generic... but it was before my time).

One can get to the machine via its IP address, but not via browsing the Microsoft Network... Macs, Linux, Windows - none can see the server in browsing.

Any idea what may be wrong? Server works fine by IP.

I had a situation where I found that /boot had filled up. So I cleared up some space, then reinstalled the most recent kernel images in case the kernels were corrupt. I used:

apt-get install --reinstall linux-2.6.32-33-server

(as one example). Now these reinstalled kernels will bomb out during reboot with a message Unable to locate init. (or similar).

I could fix this the hard way - by pulling apart the images and finding out why it doesn't work. The root volume is an LVM logical volume formatted with ext4, and the boot GRUB entries match other working entries.

However, I'm sure there's a reason that the package doesn't seem to get it right. I'd like to figure out if I'm doing something wrong with the packages, or if it is something else.

Why won't these reinstalled kernels (installed with APT) boot successfully?

I am trying to set up the admin application for a Tomcat 6.0.24 instance. None of the searches I've done turn up anything I can use. I am using this configuration for Apache 2.2.14:

Alias /manager /usr/share/tomcat6-admin/manager

<Directory "/usr/share/tomcat6-admin/manager">
        Options Indexes FollowSymLinks
        AllowOverride None
        allow from all
</Directory>

ProxyPass /manager ajp://localhost:8009/manager

In the tomcat-users.xml I have this:

<tomcat-users>
  <role rolename="tomcat"/>
  <role rolename="admin"/>
  <role rolename="operator"/>
  <role rolename="manager"/>
  <user username="admin" password="nopasswordforyou" roles="admin,tomcat,manager"/>
  <user username="operator" password="nevermind" roles="operator"/>
</tomcat-users>

I found the docs that suggested I needed manager-gui role installed and defined, but that appears to be Tomcat 7, not Tomcat 6.

The manager.xml is the default provided with Ubuntu Lucid Lynx 10.04:

<Context path="/manager" 
        docBase="/usr/share/tomcat6-admin/manager" 
        antiResourceLocking="false" privileged="true" />

If I access /manager from a web browser, I get a 404 error from Tomcat: "requested resource not available." If I access /manager/images I get the same thing. If I access /manager/401.jsp I get the actual page.

In addition, the server.xml has not only the usual Realm (UserDatabaseRealm) but also one for MySQL authentication (JDBCRealm). Investigating this showed that the role of manager was not present there for the user admin; I fixed that by doing:

INSERT USER_ROLE_DB SET USER_NAME='admin', ROLE_NAME='manager';

I restarted Tomcat, although I suspect that was overkill. No change. I don't see any errors in catalina.out or in localhost.* log files.

What am I missing? What is the interaction between the different realms? How do I get the manager application working?

Recently, I updated Ubuntu 10.04 Server LTS. The server reports itself as

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=10.04
DISTRIB_CODENAME=lucid
DISTRIB_DESCRIPTION="Ubuntu 10.04.3 LTS"

There are multiple machines here with the same install and hardware, and all work fine except for one. This one has problems during the boot process:

1. When GRUB loads, it stops at the main menu and waits for keyboard input.
2. After pressing a key, GRUB reports error: no argument specified and continues (either with or without keypress).
3. Network doesn't come up: eth0 and eth1 are redefined as eth4 and eth5, and eth5 doesn't come up. ifup eth5 works fine at the command line.

I'm still working on the last - but I suspect the first two are part of the same problem. Checking differences between a working /boot/grub/grub.cfg and this one don't turn up signifcant differences: only (previous) kernel versions and root UUIDs.

A significant difference between this and other similar (working) systems in our data center is that this system is a Dell R710; others are Dell 2950s. We've working R710s all over that are running Ubuntu Server 10.04 but none updated as currently as this one.

This problem has just recently surfaced; I've updated and rebooted three systems including this one in the last 48 hours. Only this one had problems - but only this one stopped at the GRUB menu. The others all come back automatically - so did I just miss seeing this error perhaps?

Grub is version 1.98:

| Status=Not/Inst/Cfg-files/Unpacked/Failed-cfg/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name                            Version                           Description
+++-===============================-=================================-============================================
ii  grub-common                     1.98-1ubuntu12                    GRand Unified Bootloader, version 2 (common 
ii  grub-pc                         1.98-1ubuntu12                    GRand Unified Bootloader, version 2 (PC/BIOS

I found this thread on the Ubuntu forums but it seems to be related to Natty (11.10). Some reports suggested this bug showed up in an upgrade from Lucid (10.04) to Maverick (11.04) or from Maverick to Natty; I'm only running Lucid LTS. This thread is also related to GRUB 1.99 and this version is 1.98.

The thread suggests that the line

search --no-floppy --fs-uuid --set 857d5af9-23cd-4d9b-908b-cc075e866758

should instead be

search --no-floppy --fs-uuid --set=root 857d5af9-23cd-4d9b-908b-cc075e866758

Can someone help resolve this issue? Am I just seeing this error here because GRUB stops at the menu? Why would GRUB stop at the menu on one system and not others?

UPDATE: While the actual version is 1.98 the real version of GRUB is GRUB 2; this also holds for version 1.99.

The more I think about this, the more convinced I am that other systems are probably reporting the same error but continuing on; the real problem here (with GRUB anyway) is the fact that the system stops at the menu. I can't test the other systems for comparison as they are back in use already.

UPDATE: I rebooted the machine again. The message first shows up (in bold) before the menu appears. After selecting the appropriate option, it shows up (on a clean screen) as described above with a Press any key to continue... message. After about 5 seconds, the screen clears again and booting continues (minus networking).

The GRUB2 installation reports itself as version 1.99~rc1-13ubuntu3 which I think is different from what the package reports. I'm not sure if this is just the way it is or if things have been changed.

UPDATE: More on the two versions of GRUB2 mentioned. The first is the installed package which is 1.98-1ubuntu12; the second is the version reported at the GRUB2 menu which is 1.99~rc1-13ubuntu3. According to the changelog for the GRUB2 package, version 1.99~rc1-13ubuntu3 was introduced on April 21, 2011, in Natty (note, not Lucid) and version 1.98-1ubuntu12 was introduced June 17, 2011, in Lucid Updates. This system has never had Ubuntu Natty or Oneric on it; it did however, have Red Hat Enterprise Linux 5 on it. However, the version of GRUB used by RHEL 5 (at time of writing) is 0.97-13.5 which is GRUB Legacy.

So where did version 1.99~rc1-13ubuntu3 come from? I strongly suspect that this is the source of my problems, and a complete reinstall of GRUB2 to the MBR would fix my problems. However, I've had to return the system to full operation; thus, testing will have to wait.

I can't find the answer to this anywhere: Does a module installed via APT conflict with one installed via CPAN?

Along with that question are these: Where do the installed CPAN modules live? Are they accessible globally or just to the installing user? How do you know which one? How do you know what's installed and which takes precedence?

Should CPAN even be used when modules are installed via APT?

This is the most unusual thing. I'm trying to start up mysqld with a different my.cnf (so I can have two MySQL daemons running without conflict). The file is /etc/mysql/my2.cnf but mysql won't open it.

When I run this command:

sudo -u mysql strace /usr/sbin/mysqld --defaults-file=/etc/mysql/my2.cnf

I see this in the output:

stat("/etc/mysql/my2.cnf", {st_mode=S_IFREG|0644, st_size=3574, ...}) = 0
open("/etc/mysql/my2.cnf", O_RDONLY)    = -1 EACCES (Permission denied)

However, when I change the command to:

sudo -u mysql strace cat /etc/mysql/my2.cnf > /dev/null

I see this in the output:

fstat(1, {st_mode=S_IFCHR|0666, st_rdev=makedev(1, 3), ...}) = 0
open("/etc/mysql/my2.cnf", O_RDONLY)    = 3

Same user - same kernel call - different results!

I checked file permissions - and extended permissions:

# ls -l $PWD/my2.cnf
-rw-r--r-- 1 root root 3574 2011-07-08 10:04 /etc/mysql/my2.cnf
# lsattr $PWD/my2.cnf
-----------------e- /etc/mysql/my2.cnf

Am I getting bit by some part of AppArmour or SELinux? I didn't see anything like that in the logs (including daemon.log, syslog, or messages).

How do I fix this problem?

I know about pinning packages with Apt. That's not what I want to do. Other questions have been answered with either using pinning or by using pins temporarily. I don't want to do this.

What I want to do is keep packages back the same way the kernel has been:

# apt-get upgrade
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following packages have been kept back:
  linux-generic-pae linux-headers-generic-pae linux-image-generic-pae
The following packages will be upgraded:

I want to add tomcat-* and mysql-* and sun-* to this list. In the past, there was a configuration parameter to do this. I've always thought it was something like Apt::Get::HoldPkgs or Apt::HoldPkgs but I can't find it.

I want to have these packages held from updates until I specifically request them with an apt-get install.

I found the apt-get configuration Apt::NeverAutoRemove. Will this do what I want?

Added Question: I notice that Apt::NeverAutoRemove and Apt::Never-MarkAuto-Sections (among others) are not documented so far as I can see. They're not in the manpages. Neither is aptitude::Keep-Unused-Pattern and aptitude::Get-Root-Command.

Is there any comprehensive and complete documentation for apt.conf?

How is it that Cfengine does not exist in the Ubuntu (10.04 LTS) Main Repository? I can't find a discussion of this anywhere (using Google). The only configuration management in Ubuntu Main seems to be Puppet.

I looked for a wide variety of others as well - all from Wikipedia's list of configuration management tools - and none of them are present in Ubuntu main. I looked for bcfg2, opensymbolic, radmind, smartfrog, spacewalk, staf, synctool, chef - none are present.

From my vantage point as a system administrator, I would have expected to find at least bcfg2, puppet, cfengine, and chef (as the most widely used tools).

Why is cfengine (or chef and others) not included in Ubuntu main? Why is there only one configuration management tool in Ubuntu main?

By the way - the reason this is important in the context of server administration is because Ubuntu main is fully supported by the Ubuntu team with updates and security updates; the other repositories are not.

I can't seem to find an answer to this question. There is the Linux Crypto API, and the OpenBSD Crypto Framework (OCF). I can't find anyting on Google as to whether Ubuntu supports OCF or if the Ubuntu kernel uses the Linux Crypto API.

I'm particularly interested in Ubuntu 10.04.2 LTS Server if that matters.

I'm sure I've seen a picture of the full building somewhere. The image is a blue-tinted display of the building's facade - facing into a blue sky. It's a part of the stock wallpapers in Ubuntu 10.10 (Maverick).

Can someone help me identify this building? I can't be the first - or the only - person to be interested in knowing.

I created an APT repository using deb-scanpackages, and it seemed to go well.

When I did a apt-get update on another server, the Packages.gz file was retrieved, and all seemed well - until I went to search for the packages contained in that repository (all packages are created locally).

Several recommendations suggested reprepro; I tried that. Same result - except I had to rebuild the packages with the Priority and Section lines in the control file (nothing says this anywhere). The reprepro utility also generates a complicated directory structure which required rewriting the repository entry on the requesting server.

I then found that the arch directory referenced i386 and not amd64 (which was requested by the requesting server).

Is it possible that the AMD64 system isn't seeing packages compiled for i386? Searching the *Packages files in /var/lib/apt/lists show that the only packages for i386 are those I added (the other files are for the server - Ubuntu 10.04.2 LTS).

The server the packages were built on is Ubuntu 10.04.2 LTS i686; the requesting server is x86_64.

I found some discussion at the Debian AMD64FAQ but it claims to be obsolete. It makes mention of an extended syntax for repository listings for APT, and a command dpkg-subarchitecture - neither of which work on the local AMD64 server.

Do I have to build two different sets of packages?