[metrics-team] Collector 1.1 Instance

[iwakeh]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256



On 11/19/2016 06:16 PM, Tom Ritter wrote:
> [snip] 
> 
> 
> I think this fixed it, hopefully my json file looks correct now?

Glad to hear!

> Would
> you be able to either point me to the 'json file sanity checkers' part
> of collector or list them for me?  I'd like to improve my current
> alert[0] to be more robust.

As index.json is a file listing of the instance's archive and recent folders,
there are several points to check in index.json in order to tell, if CollecTor
is doing fine

* "index_created":"2016-11-21 09:30" should be at most 2min old with the default
  setting, i.e. this tells if the index is current.
* The archives of the current month should be at most three days old (i.e. the
  time interval of the tar-script runs).  For example,
  {"path":"server-descriptors-2016-11.tar.xz","size":116953688,"last_modified":"2016-11-20 12:04"}
  is now just about one day old.  So, this tells us the archive creation worked.
* And, checking, if current files are present, e.g., for the current consensus 
  {"path":"2016-11-20-08-00-00-consensus","size":1587073,"last_modified":"2016-11-20 08:05"}
* And, as MetricsCheccker.py seems to check, if the current consensus is actually served.
  (Although, I cannot make sense of the URL there, starting with
  "https://example.com/out/relay-descriptors/consensus/".  The 'out' folder is just an
  internal folder not exposed by the http-server.  I'd replace this by:
  "https://collector.ritter.vg/recent/relay-descriptors/consensuses/<curent-date-time>-consensus".
* My collector mirror sends a mail as soon as an ERROR is logged, which would have
  alerted me if the heap-space problem had happened on my instance. 

We have an Operator wiki here [0].
Would it be ok, if I set a link to your MetricsChecker.py?  As example for others.
You're welcome to also add information there.

Kind regards,
iwakeh

[0] https://trac.torproject.org/projects/tor/wiki/org/teams/MetricsTeam/Operation

> 
> -tom
> 
> [0] https://github.com/tomrittervg/checker/blob/master/samplejobs/MetricsChecker.py
> just checks for a daily consensus file
> _______________________________________________
> metrics-team mailing list
> metrics-team at lists.torproject.org
> https://lists.torproject.org/cgi-bin/mailman/listinfo/metrics-team
> 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iF4EAREIAAYFAlgyxKIACgkQsANAxTOj/abmOwD+MRzy8hrGli68zI+FG1Jf194N
1AHaJp8w9tKC1tCW/dYA/jIs4Prg+km4wdpi4JznF2cHrfpMgiu5dFvf2OlW7gBi
=SWtE
-----END PGP SIGNATURE-----