messages in our stderr

Discussion in 'Ruby/Rails' started by cswilliams, Mar 22, 2008.

  1. cswilliams

    cswilliams New Member

    Hi I noticed in our stderr log we have many of these:
    /opt/lsws/fcgi-bin/RailsRunner.rb:23:in `accept': SIGUSR1 (SignalException)
    from /opt/lsws/fcgi-bin/RailsRunner.rb:23


    Any idea what could be causing this?

    We are running ruby-lsapi 3.0 and litespeed 3.3.4.
    I just upgraded ruby-lsapi on our machines to 3.1. Will let you know if that makes a difference.
  2. cswilliams

    cswilliams New Member

    i guess from reading some other threads, this message is normal and just due to it killing off old child processes?


    I am running into a problem that happens intermittently with our rails app. Specifically, I click Submit on our login form after putting in my username/password. Within a second or two my browser gives back a "Server unexpectedly dropped the connection" message. I don't believe the request goes through completely, since after I refresh the page, I still am not logged in and have to log in again. There's nothing interesting in any of the logs, stderr reports nothing, and the access log and rails log appear to get the request just fine. Any ideas?
  3. mistwang

    mistwang LiteSpeed Staff

    You can enable the DEBUG log, and check how the request has been processed.
  4. cswilliams

    cswilliams New Member

    ok i turned on the debug log. Here's what it had for one of the times when I got the 'server dropped connection' message from my browser:

    2008-03-29 17:00:04.710 DEBUG [*:443] New connection from 67.170.197.205:54020.
    2008-03-29 17:00:04.710 DEBUG [*:443] 1 connections accepted!
    2008-03-29 17:00:05.607 DEBUG [67.170.197.205:54020-0] [SSL] SSLAgain()!
    2008-03-29 17:00:06.480 DEBUG [67.170.197.205:54020-0] HttpIOLink::handleEvents() events=1!
    2008-03-29 17:00:06.480 DEBUG [67.170.197.205:54020-0] [SSL] SSLAgain()!
    2008-03-29 17:00:06.679 DEBUG [67.170.197.205:54020-0] [SSL] SSLAgain()!
    2008-03-29 17:00:07.679 DEBUG [67.170.197.205:54020-0] [SSL] SSLAgain()!
    2008-03-29 17:00:08.679 DEBUG [67.170.197.205:54020-0] [SSL] SSLAgain()!
    2008-03-29 17:00:09.679 DEBUG [67.170.197.205:54020-0] [SSL] SSLAgain()!
    2008-03-29 17:00:09.679 DEBUG [67.170.197.205:54020-0] Keep-alive timeout, close!
    2008-03-29 17:00:09.679 DEBUG [67.170.197.205:54020-0] Shutting down SSL ...
    2008-03-29 17:00:09.679 DEBUG [67.170.197.205:54020-0] Shutting down out-bound socket ...
    2008-03-29 17:00:10.679 DEBUG [67.170.197.205:54020-0] Shutdown time out!
    2008-03-29 17:00:10.679 DEBUG [67.170.197.205:54020-0] Close socket ...
    2008-03-29 17:00:28.016 DEBUG [idle] Idle connection timed out, close!
    2008-03-29 17:00:28.016 DEBUG [idle] [ExtConn] close()

    I thought maybe from the message the keep alive timeout was too low, so I changed it from 5 seconds to 35. But it didn't seem to affect anything.
  5. mistwang

    mistwang LiteSpeed Staff

    Do you get this error frequently?
    The keep-alive timeout definitely play a role here. 35 seconds should help.

    I have change the timeout calculation in 3.3.10, you can give it a try by changing the version number in the download link.
  6. cswilliams

    cswilliams New Member

    No, I don't get the error that frequently, although I worry if my visitors see it frequently. I have a lot of trouble repeating it. It seems to happen pretty randomly. Sometimes I don't see it for weeks.

    Today however, I kept getting the 'connection dropped' message from my browser continuously. I was uploading a large file to amazon s3 on my machine at the same time, so my connection was quite saturated. (Our lsws server is not on the same network as my machine)

    I'm convinced this had something to do with the error since once the file finished uploading, I could connect to our server just fine with no 'drop' errors. FYI, although slow, I could connect to other sites/servers including the lsws admin page of the server just fine while I was uploading this file, so I think something isn't right. In the past though, I haven't been uploading anything when I've received this 'drop' message in my browser, so I'm not sure if it's related to a slow connection, or I'm just seeing two different issues.

    I'll try your update and see if I can repeat the issue.
  7. cswilliams

    cswilliams New Member

    Well it seems to be better with the update. I tried repeating everything again. I started uploading a big file and only 1 out of 10 times did I get the drop message. Much better than before I was getting it continuously.
  8. mistwang

    mistwang LiteSpeed Staff

    What is your timeout value with the update release?
  9. cswilliams

    cswilliams New Member

    35 seconds
  10. cswilliams

    cswilliams New Member

    Actually, I'm still getting the dropped connection message on our production server. I think I finally caught the debug message:
    2008-04-05 01:47:41.410 DEBUG [10.1.0.3:38715-0] SSL_accept() failed!: error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request

    Could this be causing it?
  11. cswilliams

    cswilliams New Member

    Here's a full log for the request which exhibits the same problem, if you need it. It seems to happen randomly when I POST to our https forms. Most of the time it does work fine though, thats why I find it really strange. Hopefully it can be fixed.

    2008-04-05 01:50:54.703
    DEBUG
    [*:443] New connection from 10.1.0.2:30354.
    2008-04-05 01:50:54.703
    DEBUG
    [*:443] 1 connections accepted!
    2008-04-05 01:50:54.703
    DEBUG
    [10.1.0.2:30354-0] HttpIOLink::handleEvents() events=1!
    2008-04-05 01:50:54.703
    DEBUG
    [10.1.0.2:30354-0] [SSL] SSLAgain()!
    2008-04-05 01:50:54.703
    DEBUG
    [10.1.0.2:30354-0] SSL_accept() failed!: error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
    2008-04-05 01:50:54.703
    DEBUG
    [10.1.0.2:30354-0] Shutting down SSL ...
    2008-04-05 01:50:54.703
    DEBUG
    [10.1.0.2:30354-0] Shutting down out-bound socket ...
  12. mistwang

    mistwang LiteSpeed Staff

    Our latest 3.3.10 build should be better. You can give it a try.
  13. cswilliams

    cswilliams New Member

    I upgraded to 3.3.10, but still getting the dropped connection message occasionally when I POST to https forms on our site. The odd thing is that I cannot get it to give this error during the day, only at night. We have a lot of outbound traffic on our network (remote backup,etc.) at night, maybe that has something to do with it? Just an idea.

    Here's the latest message from the 3.3.10 version:
    2008-04-06 01:15:50.141
    DEBUG
    [*:443] New connection from 10.1.0.2:40745.
    2008-04-06 01:15:50.141
    DEBUG
    [*:443] 1 connections accepted!
    2008-04-06 01:15:50.141
    DEBUG
    [10.1.0.2:40745-0] HttpIOLink::handleEvents() events=1!
    2008-04-06 01:15:50.141
    DEBUG
    [10.1.0.2:40745-0] [SSL] SSLAgain()!
    2008-04-06 01:15:50.141
    DEBUG
    [10.1.0.2:40745-0] SSL_accept() failed!: error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
    2008-04-06 01:15:50.141
    DEBUG
    [10.1.0.2:40745-0] Shutting down SSL ...
    2008-04-06 01:15:50.141
    DEBUG
    [10.1.0.2:40745-0] Shutting down out-bound socket ...
  14. mistwang

    mistwang LiteSpeed Staff

    This message looks like http://... is used to access the https site. The first package from client is not a SSL CLIENT HELLO . It may not related to the connection reset problem.

Share This Page