3

I'm seeing a significant number of 503 errors in the logs for se-flair from attempts to load users' Stack Exchange accounts page at http://stackexchange.com/users/[...]?tab=accounts (where [...] may be an old-style association ID hash or a new-style user ID number) over the past 24 hours, where I hadn't seen many, if any, such errors before. It's not happening 100% of the time and the infrastructure on Google App Engine isn't giving me much detail on the failed URL loading beyond that it's a 503 error:

urlfetch of http://stackexchange.com/users/110007?tab=accounts failed: 503
{'connection': 'close', 'via': 'HTTP/1.1 GWA', 'content-type': 'text/html', 'x-google-cache-control': 'remote-fetch', 'cache-control': 'no-cache'}

Is this something transient with Stack Exchange in general or something to do with my app in particular?

Edit: The issue has continued for over 7 days now. It's not affecting a huge portion of requests, but that's probably because once the accounts-tab URL has successfully loaded, the data is cached for 8 hours. Below is the graph of requests and errors over the past 30 days. Note that the orange error line was basically flat until 7-8 days ago (and where it wasn't, it wasn't caused by this particular 503 error). Over the past 7-8 days, this particular 503 error represents more than 95% of the errors in that orange line.

request/error graph

Improve this question
2
  • Out of curiosity though, why do you need to load that page?
    – Tim Stone
    Commented Feb 2, 2014 at 3:33
  • 1
    @TimStone: It was (and I think still is) the only way to get complete information about all of a user's accounts including Area 51 (the /2.1/users/{ids}/associated endpoint wasn't available when I started and doesn't seem to include Area 51). se-flair should be caching the retrieved data for up to 8 hours for each individual user (and thus not hammering the URL), though that can wind up shorter if the Google App Engine instance doesn't stay alive (going down to 0 active instances seems to clear the cache).
    – Isaac
    Commented Feb 2, 2014 at 4:11

1 Answer 1

Reset to default
2

We had two sites change names in the past 7 days. These aren't exactly painless and it will probably settle back to normal in the next 24 hours.

Improve this answer
1
  • The site-renames somehow temporarily cause the accounts tab of the Stack Exchange profile to yield a 503 error? Would the errors be specific to users who had accounts on the renamed sites? If so, could you specify which sites were renamed so that I could check that the users for whom the errors were occurring correlated?
    – Isaac
    Commented Feb 8, 2014 at 19:37

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .