Requester Doesn't Recover From EOFError

[josephholsten]

We got this issue in hoptoad:

Error Message: EOFError: end of file reached

Where: /usr/ruby1.8.7/lib/ruby/1.8/net/protocol.rb, line 135

Backtrace Summary:

/usr/ruby1.8.7/lib/ruby/1.8/net/protocol.rb:135:in `sysread'
/usr/ruby1.8.7/lib/ruby/1.8/net/protocol.rb:135:in `rbuf_fill'
/usr/ruby1.8.7/lib/ruby/1.8/timeout.rb:62:in `timeout'
/usr/ruby1.8.7/lib/ruby/1.8/timeout.rb:93:in `timeout'
/usr/ruby1.8.7/lib/ruby/1.8/net/protocol.rb:134:in `rbuf_fill'
/usr/ruby1.8.7/lib/ruby/1.8/net/protocol.rb:116:in `readuntil'
/usr/ruby1.8.7/lib/ruby/1.8/net/protocol.rb:126:in `readline'
/usr/ruby1.8.7/lib/ruby/1.8/net/http.rb:2026:in `read_status_line'
/usr/ruby1.8.7/lib/ruby/1.8/net/http.rb:2015:in `read_new'
/usr/ruby1.8.7/lib/ruby/1.8/net/http.rb:1051:in `__request__'
[PROJECT_ROOT]/.bundle/gems/ruby/1.8/gems/rest-client-1.6.1/lib/restclient/net_http_ext.rb:17:in `request_without_newrelic_trace'
/usr/ruby1.8.7/lib/ruby/1.8/net/http.rb:772:in `get'
[PROJECT_ROOT]/.bundle/gems/ruby/1.8/bundler/gems/rets4r-1ef13955a443/lib/rets4r/client/requester.rb:94:in `request'
/usr/ruby1.8.7/lib/ruby/1.8/net/http.rb:543:in `start'
[PROJECT_ROOT]/.bundle/gems/ruby/1.8/bundler/gems/rets4r-1ef13955a443/lib/rets4r/client/requester.rb:77:in `request'
[PROJECT_ROOT]/.bundle/gems/ruby/1.8/bundler/gems/rets4r-1ef13955a443/lib/rets4r/client.rb:461:in `request'
[PROJECT_ROOT]/.bundle/gems/ruby/1.8/bundler/gems/rets4r-1ef13955a443/lib/rets4r/client.rb:205:in `login'
[PROJECT_ROOT]/lib/rets_service/search_response.rb:43:in `each'
[PROJECT_ROOT]/lib/job/photo_sync/two_pass_strategy.rb:28:in `map'
[PROJECT_ROOT]/lib/job/photo_sync/two_pass_strategy.rb:28:in `remove_orphaned_photos'
[PROJECT_ROOT]/lib/job/photo_sync/two_pass_strategy.rb:10:in `sync_photos'
[PROJECT_ROOT]/lib/job/photo_sync.rb:7:in `perform_without_newrelic_transaction_trace'

[kidbrax]

Thought I commented on this issue recently but I guess not. Anyway, I am also getting this but only on one specific board. I am on ruby 1.9.2. Has anyone looked any further into this? I am stumped.

[josephholsten]

Is it at all reproducible? Maybe we should add some logging to see if we can correlate things?

[kidbrax]

It looks like it's an error with right_http_connection.
flexera-public/right_http_connection#5

[kidbrax]

So it turns out I was trying to use the RETS offset parameter and this was causing the issue. Some boards allow offset and others don't. Still appears to be a bug since it should have returned a better error instead of reaching EOF. But the bug is not in Rets4r. Anyway, the workaround was to not use offset to avoid the EOF error but also extend the read timeout so that the entire query could be returned before it times out when not using offset.

[kidbrax]

So thought I had this fixed but unfortunately this issue has raised its head again.

I did some work with rest-client and as you pointed out it has net/http at core so I got the same problem. But I can reproduce this bug every time with a certain board I am working with. If I gave you the credentials, would you be willing to take a look at it?

I can't confirm this yet, but it seems that it may be related to the RETS action urls. This board has the action url but is unable to process the request for that url. If I skip the action url, then I get the error above.