Anentropic's questions

docker-machine provides an upgrade command, however it only upgrades to the latest stable version

I want to install 1.10RC2 on my docker machines and I am uncertain the steps to do this

UPDATE

On the docker releases page it shows this command:
curl -fsSL https://test.docker.com/ | sh

I thought I'd try that out. I ssh'd in via docker-machine ssh swarm-master and ran it.

It looked promising... running it appears to be an install/upgrade script, it did a bunch of stuff. Afterwards:

root@swarm-master:~# docker version
Client:
 Version:      1.10.0-rc2
 API version:  1.22
 Go version:   go1.5.3
 Git commit:   c1cdc6e
 Built:        Wed Jan 27 22:14:06 2016
 OS/Arch:      linux/amd64
Cannot connect to the Docker daemon. Is the docker daemon running on this host?

Seems to have upgraded, but daemon not running.

https://docs.docker.com/engine/articles/systemd/ shows a couple of ways. I tried:

$ sudo systemctl start docker

and

$ sudo service docker start

In both cases the command returns without output but hasn't seemed to have done anything, the daemon is still not running.

My host is Debian 8.

I can do docker daemon and it runs in foreground... doing so for the first time it went through and migrated all my images... seems to work fine but

So my question is now - how to properly (as in: same way that docker-machine would do it) run the daemon?

Docs say:

Under this configuration, when the container consumes memory more than 200M and less than 500M, the next system memory reclaim attempts to shrink container memory below 200M.

But what does this actually mean?

Let's say I have a container running Solr, and:
docker run -m 1G --memory-reservation 768M solr

What effect does it have on the Solr process inside the container when it starts using 800M memory? Nothing?

Is it only if/when memory usage subsequently drops below 768M... this flag allows docker to reclaim the memory from the solr container? ...otherwise the container would continue to consume whatever the high-water memory usage was?

How does this relate to swarm scheduling? If swarm always allocates 1G for the container because of -m then is --memory-reservation not useful in this context?

I encountered this problem with my Postfix install:

https://www.virtualmin.com/node/23909

i.e. I was getting the error:

postfix/smtp[130]: fatal: unknown service: smtp/tcp

and like at the url above, the solution was to manually copy /etc/services and /etc/resolv.conf into /var/spool/postfix/etc

My impression is this "just started happening" and previously wasn't a problem, however I have only been making minor seemingly unrelated changes to postfix main.cf so I don't understand why.

The Postfix installation tutorials I was following don't mention anything about having to manually copy those files... this feels like something Postfix ought to do by itself and isn't for some reason.

Is there some underlying misconfiguration that would cause this?

I am trying to configure Postfix for the first time. I do not need mailboxes, I want only virtual aliases, forwarding [email protected] --> [email protected]

My Postfix is running in a Docker container on a Digital Ocean droplet.

I have got as far as:

$ postalias -q [email protected]
[email protected]

...from inside the container, i.e. my /etc/postfix/virtual file is working.

Also, from outside the container on the droplet:

telnet example.com 25
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
220 example.com ESMTP Postfix (Ubuntu)

But from my own computer:

$ telnet example.com 25
Trying <droplet IP>...
telnet: connect to address <droplet IP>: Operation timed out
telnet: Unable to connect to remote host

I think this is expected and correct as a result of the mynetworks config setting (see below) which is as recommended by Digital Ocean - I don't want to host an 'open' SMTP relay.

I have an MX record for example.com. set up and I can ping example.com just fine and access websites on it. Also (from my computer):

$ host -t mx example.com
example.com mail is handled by 1 example.com.

So that looks right.

But if I send a test message to [email protected] nothing comes through and I don't see anything in Postfix logs either... I can't tell at what point it failed.

I am uncertain if the problem is in my Postfix configuration or in the routing into the container.

The container exposes port 25 (only) and is run via Fig with

ports:
  - "25:25"

From shell in the droplet:

$ netstat -tulpn | grep 25
tcp6       0      0 :::25                   :::*                    LISTEN      10680/docker-proxy

My /etc/postfix/main.cf has this in it:

smtpd_banner = $myhostname ESMTP $mail_name (Ubuntu)
biff = no

# appending .domain is the MUA's job.
append_dot_mydomain = no

# Uncomment the next line to generate "delayed mail" warnings
#delay_warning_time = 4h

readme_directory = no

# TLS parameters
smtpd_tls_cert_file=/etc/ssl/certs/ssl-cert-snakeoil.pem
smtpd_tls_key_file=/etc/ssl/private/ssl-cert-snakeoil.key
smtpd_use_tls=yes
smtpd_tls_session_cache_database = btree:${data_directory}/smtpd_scache
smtp_tls_session_cache_database = btree:${data_directory}/smtp_scache

smtpd_relay_restrictions = permit_mynetworks permit_sasl_authenticated defer_unauth_destination
myhostname = example.com
alias_maps = hash:/etc/aliases
alias_database = hash:/etc/aliases
mydestination = /etc/mailname, <container id>, localhost.localdomain, localhost, example.com
relayhost =
mynetworks = 127.0.0.0/8 [::ffff:127.0.0.0]/104 [::1]/128
mailbox_size_limit = 0
recipient_delimiter = +
inet_interfaces = all
virtual_alias_domains = example.com
virtual_alias_maps = hash:/etc/postfix/virtual
inet_protocols = ipv4

I'm not really clear of the distinction between myhostname (which was originally set to <container id>) mydestination and virtual_alias_domains

Updated

with output from http://mxtoolbox.com/SuperTool.aspx

Connecting to <server IP>

220 example.com ESMTP Postfix (Ubuntu) [733 ms]
EHLO MXTB-PWS3.mxtoolbox.com
250-example.com
250-PIPELINING
250-SIZE 10240000
250-VRFY
250-ETRN
250-STARTTLS
250-ENHANCEDSTATUSCODES
250-8BITMIME
250 DSN [714 ms]
MAIL FROM: <[email protected]>
250 2.1.0 Ok [722 ms]
RCPT TO: <[email protected]>
454 4.7.1 <[email protected]>: Relay access denied [715 ms]

MXTB-PWS3v2 4006ms

Update
As @masegaloeh helped me discover, my postfix server was basically working. I had two problems that confused things though:

1. Due to mistake in my Dockerfile I had /var/log/mail.log owned by root user... this is why it stayed empty. I didn't see any errors about it but basically rsyslog couldn't write to it. A chown syslog:adm /var/log/mail.log step fixed that, and I was able to see that indeed postfix was handling and forwarding mail to the alias.

2. I mistakenly believed I was able to telnet to other servers on port 25 from my laptop, just because I thought I'd done stuff in the past that would need that to work. But actually I can't. However I was able to telnet example.com 25 from another server so, again, stuff was actually working.

3. Sending mail to the alias from another server works, and comes through to my Gmail destination address.

4. It seems my problem is actually with Gmail... when I send the message to [email protected] from my Gmail account it does not show up. I since tried with aliases I had setup on another hosting... some work and some don't... leading to:

Conclusion:
it seems that Gmail will only accept mail for aliases that are configured under Settings > Accounts and Import > Send mail as ...unfortunately Gmail now requires you to specify a 3rd party SMTP server for it when setting up a new one, so it looks I will have to get to grips with TLS etc in my postfix install.

I have a Dockerfile with:

ENTRYPOINT ["uwsgi", "--ini /home/docker/app/uwsgi_app.ini"]

(no CMD)

When I run that uwsgi rightly complains:

uWSGI running as root, you can use --uid/--gid/--chroot options
*** WARNING: you are running uWSGI as root !!! (use the --uid flag) ***

Now, I thought that I can append arguments to the entrypoint via docker run like so:

$ docker run -itP uwsgi_app:0.1 --uid=docker

However uwsgi still complains with the same error, it seems the arg has not been appended.

If I override the entrypoint then arg gets added:

$ docker run -itP --entrypoint bash uwsgi_app:0.1 --uid=docker
bash: --uid=docker: invalid option

... I wondered if it was uwsgi's --ini option ignoring other args, but it's not that:

$ docker run -itP --entrypoint bash uwsgi_app:0.1
[ root@88941de25b1f:/home/docker ]$ uwsgi --ini /home/docker/app/uwsgi_app.ini --uid=docker
user docker not found.

(okay, an error, but shows the arg was accepted)

I am mostly using 'git-bash' which came with the Windows install of Git vcs. I think it is the same one that comes with Cygwin. It works fine technically but the UI sucks:

• have to right-click the title bar and go to Properties to change the window width
• most annoyingly... copy and paste and highlight with the mouse are equally cumbersome

In contrast, the Terminal app that comes with OS X manages these with aplomb and makes it so much more comfortable to work with. You can even drag and drop a file on it and it will paste the file path in at your cursor!

I have also tried:

• http://sourceforge.net/projects/win-bash
• http://www.steve.org.uk/Software/bash/
• http://www.hamiltonlabs.com/cshell.htm

None of these do copy and paste of text without cumbersome right-clicking.

I am specifically looking for a Unix-flavoured shell in Windows so I don't have to use different shells between my home dev (Windows), the live server (Linux) or dev at the office (Mac).

Yes I have Googled and haven't found one yet...