Timing out on getsentry.com (Europe Region)

[troelskn]

Hi,

I've signed up at getsentry.com and installed the gem from source, e.g. in Gemfile I have:

gem 'sentry-raven', :git => "https://github.com/getsentry/raven-ruby.git"

I run the test command bundle exec rake raven:test and get the following error:

I, [2013-10-11T22:28:24.700397 #7352]  INFO -- : ** [Raven] Raven 0.6.0 ready to catch errors
Client configuration:
-> server: https://app.getsentry.com
-> project_id: 12345
-> public_key: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
-> secret_key: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Sending a test event:
-> ** [Raven] Unable to record event with remote Sentry server (Faraday::Error::TimeoutError - execution expired)

An error occurred while attempting to send the event.

It happens pretty fast, so I don't think there is an actual timeout happening.

Any ideas?

[dcramer]

I haven't heard of a case where it immediately times out, though we do have very low timeouts (usually 1s) by default.

If you change the https to http do you have the same issue?

[troelskn]

Changing from https to http resolves the issue.

[nateberkopec]

Hmmm...if the addt'l overhead caused by the HTTPS handshake/protocol is too much, maybe 1s is too fast? Usually, if you need HTTPS, you don't really have a choice, so "just switch to https" wouldn't be a resolution here.

[dcramer]

It shouldn't be that much overhead. I'm wondering if it was a blip in the system or SSL connection is just failing in this case.

[troelskn]

Re-enabling https brings the problem back - so it's definitely not just a random fluke. I tried from home and from work, so it's not a net working issue. Could be something on my machine of course, but I don't otherwise have problems with ssl.

[dcramer]

Could you dump the results of:

time curl -I https://app.getsentry.com

I want to confirm that this isn't general network latency

On Tue, Oct 15, 2013 at 8:07 AM, Troels Knak-Nielsen
notifications@github.com wrote:

Re-enabling https brings the problem back - so it's definitely not just a random fluke. I tried from home and from work, so it's not a net working issue. Could be something on my machine of course, but I don't otherwise have problems with ssl.

Reply to this email directly or view it on GitHub:
https://github.com/getsentry/raven-ruby/issues/135#issuecomment-26313252

[troelskn]

Sure:

$ time curl -I https://app.getsentry.com 
HTTP/1.1 302 FOUND
Server: nginx
Date: Tue, 15 Oct 2013 11:50:24 GMT
Content-Type: text/html; charset=utf-8
Connection: keep-alive
Vary: Accept-Language, Cookie
Location: https://app.getsentry.com/login/
Content-Language: en-us
Set-Cookie: sessionid="gAJ9cQFVBV9uZXh0cQJYAQAAAC9zLg:1VW38q:4vKqfv8F-5dGoC1bWBFvBHRCGNA"; Domain=.getsentry.com; expires=Tue, 29-Oct-2013 11:50:24 GMT; httponly; Max-Age=1209600; Path=/; secure
Strict-Transport-Security: max-age=2592000


real    0m1.933s
user    0m0.019s
sys 0m0.015s

[dcramer]

What region are you running that from?

On Tue, Oct 15, 2013 at 12:51 PM, Troels Knak-Nielsen
notifications@github.com wrote:

Sure:

$ time curl -I https://app.getsentry.com 
HTTP/1.1 302 FOUND
Server: nginx
Date: Tue, 15 Oct 2013 11:50:24 GMT
Content-Type: text/html; charset=utf-8
Connection: keep-alive
Vary: Accept-Language, Cookie
Location: https://app.getsentry.com/login/
Content-Language: en-us
Set-Cookie: sessionid="gAJ9cQFVBV9uZXh0cQJYAQAAAC9zLg:1VW38q:4vKqfv8F-5dGoC1bWBFvBHRCGNA"; Domain=.getsentry.com; expires=Tue, 29-Oct-2013 11:50:24 GMT; httponly; Max-Age=1209600; Path=/; secure
Strict-Transport-Security: max-age=2592000
real  0m1.933s
user  0m0.019s
sys   0m0.015s

---

Reply to this email directly or view it on GitHub:
https://github.com/getsentry/raven-ruby/issues/135#issuecomment-26327750

[troelskn]

Europe (Denmark).

[longzhi]

I encountered the same problem, both http and https. I am in China

[troelskn]

Just for info - I retried this just now, and the problem isn't present any more. E.g. if I enable https, it goes through. I don't know what might have changed since, so can't advise on any fix, but I just ran it over plain http, which has worked fine (My app doesn't have any very sensitive data, so not too afraid of a leak).

[rastkojokic]

I can report the same issue... Yesterday I was not experiencing any problems while sending data to Sentry... Today I get the same error as @troelskn .

** [Raven] Raven HTTP Transport connecting to https://app.getsentry.com
** [Raven] Unable to record event with remote Sentry server (Faraday::ConnectionFailed - execution expired)
...

Any ideas what is going on?

p.s. - I am from Europe (Serbia)

[dcramer]

@rastkojokic It's really hard to say. It's sluggish networking from [somewhere], but I think its out of our control as we dont have any POPs yet (our servers are all in Texas). Definitely nothing thats different on our end as far as I can see. You may consider increasing the timeout (or ideally using the async stuff).

We may prioritize rolling out our local agent which will help alleviate any of these concerns

[IntractableQuery]

I just signed up for Sentry and was having the same timeouts when attempting rake raven:test

Switching from HTTPS to HTTP did not help.

I saw you expose faraday's timeouts (in seconds), e.g.

    config.timeout = 10
    config.open_timeout = 10

In this case, it was open_timeout that fixed the problem. I turned on async event delivery per the Advanced Configuration instructions, so hopefully this avoids locking up web request threads if I really am hitting unusually long open attempts (or does reporting even happen in the web request?).

EDIT: async doesn't seem to work reliably, turned that back off -- I assume this is not yet a supported feature