Site https://docs.pylonsproject.org is not working

[Marcin Raczyński]

Hello,

Can anybody bring https://docs.pylonsproject.org back to live?

The site is not working now:

504 Gateway Time-out

nginx/1.12.0

[Marcin Raczyński]

It is working now.

[Christoph Zwerschke]

Am 29.06.2017 um 15:07 schrieb Marcin Raczyński:
> It is working now.

see also https://github.com/rtfd/readthedocs.org/issues/2981

[Steve Piercy]

On 6/29/17 at 3:33 PM, ci...@online.de (Christoph Zwerschke) pronounced:

>Am 29.06.2017 um 15:07 schrieb Marcin Raczyński:
>>It is working now.
>
>see also https://github.com/rtfd/readthedocs.org/issues/2981

What can we do when RTD has an outage?

I have sudo access on the server now. I configured nginx to
proxy and cache 'docs.pylonsproject.org'. Evidently that was
not sufficient to handle an outage at RTD.

Any suggestions appreciated.

--steve

------------------------
Steve Piercy, Soquel, CA

[Jonathan Vanasco]

On Thursday, June 29, 2017 at 2:55:55 PM UTC-4, Steve Piercy wrote:

What can we do when RTD has an outage?

You may be able to minimize an outage by using a free CloudFlare or similar account.  Usually you can keep something in their cache for long-enough to miss periodic downtime.

Another option is to detect the issue have DNS switch to a failover.  It looks like the dns is handled by linode's nameservers, and they have an API.

[Bert JW Regeer]

On Jun 29, 2017, at 13:34, Jonathan Vanasco <jona...@findmeon.com> wrote:

On Thursday, June 29, 2017 at 2:55:55 PM UTC-4, Steve Piercy wrote:

What can we do when RTD has an outage?

You may be able to minimize an outage by using a free CloudFlare or similar account.  Usually you can keep something in their cache for long-enough to miss periodic downtime.

We explicitly decided against using CloudFlare, and using CloudFlare does not work with Read the Docs which requires you send a particular HTTP header when you proxy requests to them for their SSL setup.

Another option is to detect the issue have DNS switch to a failover.  It looks like the dns is handled by linode's nameservers, and they have an API.

This wouldn’t have solved anything either. Where are we going to switch DNS to?

--
You received this message because you are subscribed to the Google Groups "pylons-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pylons-discus...@googlegroups.com.
To post to this group, send email to pylons-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pylons-discuss/8942dba2-14bb-4ae7-ba39-3627359e8d4f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Bert JW Regeer]

> On Jun 29, 2017, at 12:55, Steve Piercy <steve.pi...@gmail.com> wrote:
>
> On 6/29/17 at 3:33 PM, ci...@online.de (Christoph Zwerschke) pronounced:
>
>> Am 29.06.2017 um 15:07 schrieb Marcin Raczyński:
>>> It is working now.
>>
>> see also https://github.com/rtfd/readthedocs.org/issues/2981
>
> What can we do when RTD has an outage?
>
> I have sudo access on the server now. I configured nginx to proxy and cache 'docs.pylonsproject.org'. Evidently that was not sufficient to handle an outage at RTD.

proxy_cache_use_stale error timeout invalid_header updating;

Add this, so that if there is a failure to retrieve the document from the remote host, and we have a cached copy of the doc, we serve up the “stale" document instead.

>
> Any suggestions appreciated.
>
> --steve
>
> ------------------------
> Steve Piercy, Soquel, CA
>

> --
> You received this message because you are subscribed to the Google Groups "pylons-discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to pylons-discus...@googlegroups.com.
> To post to this group, send email to pylons-...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/pylons-discuss/r473Ps-10125i-4077EE79F32B4402BDD67287987AB465%40Steves-iMac.local.

[Steve Piercy]

On 6/29/17 at 3:33 PM, xist...@0x58.com (Bert JW Regeer) pronounced:

>>On Jun 29, 2017, at 12:55, Steve Piercy <steve.pi...@gmail.com> wrote:
>>
>>What can we do when RTD has an outage?
>

>proxy_cache_use_stale error timeout invalid_header updating;
>
>Add this, so that if there is a failure to retrieve the
>document from the remote host, and we have a cached copy of the
>doc, we serve up the “stale" document instead.

Suh-weet! That led me to this excellent article.
https://www.nginx.com/blog/mitigating-thundering-herd-problem-pbs-nginx/

[Steve Piercy]

On 6/29/17 at 3:33 PM, xist...@0x58.com (Bert JW Regeer) pronounced:

>proxy_cache_use_stale error timeout invalid_header updating;

FTR, I had set this previously:

## caching settings per https://www.nginx.com/blog/nginx-caching-guide/
proxy_cache_use_stale error timeout updating http_500 http_502
http_503 http_504;

I inserted the `invalid_header` option, which had been omitted.

I think we've previously discussed, too, putting the Pylons
Project server config under version control and automate
deployment. Maybe it's time to revisit that topic?