Why do I constantly have to update my apt cache?

Question

When I want to install a new package, I'm usually greeted with a response like:

oliver@cloud:~$ sudo apt-get install unison
Reading package lists... Done
Building dependency tree
Reading state information... Done
E: Unable to locate package unison

Already knowing this pattern, I continue:

oliver@cloud:~$ sudo apt-get update
Hit http://downloads-distro.mongodb.org dist Release.gpg
Ign http://downloads-distro.mongodb.org/repo/debian-sysvinit/ dist/10gen Translation-en
Ign http://downloads-distro.mongodb.org/repo/debian-sysvinit/ dist/10gen Translation-en_US
Hit http://downloads-distro.mongodb.org dist Release
Ign http://downloads-distro.mongodb.org dist/10gen amd64 Packages
Ign http://downloads-distro.mongodb.org dist/10gen amd64 Packages
Ign http://http.debian.net/debian/ squeeze/contrib Translation-en_US
Ign http://http.debian.net/debian/ squeeze/main Translation-en_US
Ign http://http.debian.net/debian/ squeeze/non-free Translation-en_US
...

oliver@cloud:~$ sudo apt-get install unison
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following NEW packages will be installed:
  unison
0 upgraded, 1 newly installed, 0 to remove and 1 not upgraded.
Need to get 693 kB of archives.

If I want to install a new package tomorrow, the server will not know about it until I update again. I've only noticed this on Debian servers and I'm not aware of any adjustments that would cause this.

I understand the need for the local cache, my problem is that it seems to be reset whenever I want to install something new. I would expect the cache to stay populated for longer periods of time. Usually, when I would want to install something new the next day, apt-get will tell me it's unable to locate the package.

On this server, cron-apt is installed. I must assume it's installed by default. cron-apt is set to run nightly at 4:00. It seems to be responsible for my lost cache:

oliver@cloud:/var/lib/apt/lists$ sudo apt-get update
...
Fetched 12.1 MB in 11s (1,023 kB/s)
Reading package lists... Done
oliver@cloud:/var/lib/apt/lists$ ls /var/lib/apt/lists/ | wc -l
24
oliver@cloud:/var/lib/apt/lists$ sudo /usr/sbin/cron-apt
oliver@cloud:/var/lib/apt/lists$ ls /var/lib/apt/lists/ | wc -l
8

I'm not really sure why this is happening, because, from what I can tell, one of the jobs of cron-apt is to actually invoke apt-get update:

oliver@cloud:/var/lib/apt/lists$ cat /etc/cron-apt/action.d/0-update
update -o quiet=2

Answer 1

Your question is a little vague. If the question is "why do I have to run apt-get update" the answer is that the updating of apt can be a time consuming process. Doing it every time I had to install a package would add considerably more time. Usually updating the cache once a day (ideally, in an automated manner) is more than sufficient.

"If I want to install a new package tomorrow, the server will not know about it until I update again. I've only noticed this on Debian servers and I'm not aware of any adjustments that would cause this."

I'm fuzzy on this part. You're saying that if a new version of unison is available tomorrow, how do you know? Again, if you have apt-get update running at least once a day (say, via a cron script) then apt will learn it has new packages. That said, it's not going to automatically tell you that new packages are available; this is with good reasons, since upgrading packages can occasionally cause system stability issues, and the user is necessary to resolve them.

Generally speaking, on a Debian based system, incremental newer versions of software isn't necessarily better. I generally don't care about minor version upgrades except when there's a security issue, or if there's a feature in a new version that I know I need.

Answer 2

Usually it just means that exact version, information about which you fetched with last update, is no longer available on Debian servers as it was replaced by new one. But it shouldn't happen that much on Squeeze (even considering security updates). Other option is that some weird script (I don't know of any such one) periodically cleans your /var/lib/apt/lists directory that caches packages lists available on servers (and which apt-get update fetches). Unless you are following unstable or even experimental on your servers (which seems little weird) in which case packages are updated quite often, so cached lists are getting outdated pretty fast. So my advice would be to train your muscle memory to call apt-get update first, each time you are starting some installation session.

Answer 3

To avoid manually updating your package list, you could automate this task every night with a cron job. Use 'crontab -e' and enter the following line.

0 23 * * * apt-get update

Answer 4

Of course, you know you don't have to update until you need it.....What is happening is that the part of the catalog of information gets transmitted locally to you computer to make apt-get hopefully a little faster. Instead of multiple requests to the remote server for the entire catalog, it only makes one and then updates it as necessary. Again this process does make it faster for you, allows you to see all the packages currently available, and takes the load off of all the people simply trying to look at the catalog by loading it only once locally.

If you are only doing one package during a large period of time, it probably does appear to be a giantic time waster...but its implementation really helps everybody.