Re: [Savannah-hackers-public] migration of savannah's "download" server

[Assaf Gordon]

Some updates:

On Wed, Feb 22, 2017 at 11:53:18PM +0000, Assaf Gordon wrote:
> For the next few hours (starting at 22-Feb-2017 18:45:00 EST),
> Savannah's downloads and gnu's ftp-mirror will be
> redirected to our new servers.

So far all seems ok, not special problems observed.

few minor things:

1.
To ease troubleshooting, I've separated the nginx logs for each service.
These could be consolidated later if needed (but see ftpmirror stats 
below).
/var/log/nginx/ftpmirror.{access,error}.log
/var/log/nginx/download.{access,error}.log

2.
I see many request to a http://download.savannah.gnu.org/webalizer/ ,
for really old reports:
===
X.X.X.X - - [22/Feb/2017:21:22:21 -0500] "GET /webalizer/ HTTP/1.1" 404 163 "$query.link" 
"Mozilla/5.0 (compatible; YandexImages/3.0; +http://yandex.com/bots)"
X.X.X.X - - [22/Feb/2017:21:29:02 -0500] "GET /webalizer/usage_200902.html HTTP/1.0" 404 591 
"http://download.savannah.gnu.org/"; "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 
(KHTML, like Gecko) Chrome/39.0.2171.99 Safari/537.36"
X.X.X.X - - [22/Feb/2017:21:37:53 -0500] "GET /webalizer/usage_201302.html HTTP/1.0" 404 189 
"http://cheapessays.win"; "Mozilla/5.0 (Windows NT 5.1; rv:43.0) Gecko/20100101 
Firefox/43.0"
X.X.X.X - - [22/Feb/2017:21:37:56 -0500] "GET /webalizer/usage_201502.html HTTP/1.0" 404 591 
"https://forward-hdd.ru/burovoj-instrument/instrument-forward/"; "Mozilla/5.0 (Windows NT 6.1; 
WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/47.0.2526.111 Safari/537.36"
X.X.X.X - - [22/Feb/2017:21:48:12 -0500] "GET /webalizer/usage_201506.html HTTP/1.0" 404 591 
"http://ventolin365.accountant"; "Mozilla/5.0 (Windows NT 6.3; WOW64) AppleWebKit/537.36 
(KHTML, like Gecko) Chrome/48.0.2564.109 Safari/537.36"
===
No clue where these are coming from and why these specific dates (they
repeat many times in the logs). To see them:
$ grep weba /var/log/nginx/download.access.log


3.
/var/log/daemon.log shows repeated 'named' errors.
Is that IPv6 problem, or something else?
===
Feb 22 20:00:09 download0 named[995]: error (network unreachable) resolving 
'X.X.X.X.in-addr.arpa/PTR/IN': 2001:XXX:XX::5#53
Feb 22 20:00:09 download0 named[995]: error (network unreachable) resolving 
'ns1.xxxxxx.net/A/IN': 2001:470:XXXX:XXX::2#53
Feb 22 20:00:09 download0 named[995]: error (network unreachable) resolving 
'ns1.xxxxxx.net/AAAA/IN': 2001:470:XXXX:XXX::2#53
===


4.
ftpmirror processing now allows us to keep some statistics
about gnu package downloads.

They are available here:
 https://download.savannah.gnu.org/ftpmirror-stats/

For example to see which GCC versions are most downloaded:
 
https://download.savannah.gnu.org/ftpmirror-stats/ftpmirror-stats-packages.html#gcc

The statistics are collected every 2 hours,
see /etc/cron.d/sv-ftpmirror-stats
and /opt/savannah/bin/collect-ftpmirror-stats.sh .

If it turns out to consume too many resources (all past logs, etc.),
we can reduce the frequency or limit the number of logs checked.


5.
The previous 'download' server is now 'olddownload.savannah.gnu.org'
at 208.118.235.73 .


6.
last fun tidbit:
For some unclear reason, http://ftpmirror.gnu.org/emacs/windows is one
of the most accessed URL (and it's not an actual download, it's a
directory listing). See here:
   https://download.savannah.gnu.org/ftpmirror-stats/ftpmirror-stats-recent.txt

It is accessed from many different IPs, most (perhaps all)
are from a windows-based user-agents.
Few have HTTP-referer of 'http://www.gnu.org/software/emacs/download.html',
but many have empty referer.
If other admins want to explore that, try:
   # grep -w "emacs/windows" /var/log/nginx/ftpmirror.access.log | wc -l
   89


to be continued,
- assaf