# Strange "long polling" problem **URL:** \ **Category:** NGINX\ **Created:** [December 2, 2011, 12:29am UTC](https://www.ruby-forum.com/t/strange-long-polling-problem/214442 "2011-12-02T00:29:07Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![dubstep](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/cdc98d/32.png) [@dubstep](https://www.ruby-forum.com/u/dubstep)\ **Post date:** [December 2, 2011, 12:29am UTC](https://www.ruby-forum.com/t/strange-long-polling-problem/214442/1 "2011-12-02T00:29:07Z")
Hi All, I am having a problem with my Chatroom, that uses the long polling module linked with nginx version 1.0.6 First the structure. There are two http conversations going on with each client. GETs, created by “on-load” and receiving the previous message, are issued to the long polling “activity” address. When the data arrives, it updates the screen, as well as triggering the next “GET”. Note - there is a small time gap between one message being sent and the next being requested. POSTS are triggered by user events. These send a separate XMLHTTPRequest to a php program - handlechat.php handlechat.php writes the relevant stuff to the database, and then issues a “curl” to write data to the “publish” address, and replies with an empty “200” reply. It works 99.9% of the time. Which is not quite 100%. ☹ When it fails The client’s GET is outstanding (reported by chrome) The database has the missed message The curl write to the publish address replies with first queued messages: 1 last requested: 12 sec. ago (-1=never) active subscribers: 2 then queued messages: 1 last requested: 13 sec. ago (-1=never) active subscribers: 2 I know both CURLs worked. The “missing” message is received by the other user. It just is not received by the failing user. If the failing user refreshes the screen, thus re-initialising everything he getst the missing message twice, once from the log and once from the stalled message. As part of my testing, I tried INCREASING the delay between getting an update and requesting the next. This made the problems much much worse. It appears that messages posted after a reply is sent, and before the next request is received are not being sent on. Note - I return the etag in the header, so I think they should be. What is the correct method to over-come this problem? Regards Ian – Ian H. 31 Sheerwater, Northampton NN3 5HU, Tel: 01604 513875 Preparing eBooks for Kindle and ePub to give the best reader experience. ---
**Author:** ![dcian](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/91b2a8/32.png) [@dcian](https://www.ruby-forum.com/u/dcian)\ **Post date:** [December 2, 2011, 11:49pm UTC](https://www.ruby-forum.com/t/strange-long-polling-problem/214442/2 "2011-12-02T23:49:17Z")
Hi all, Can someone who is using the long polling module (either version .69 or .692) please respond and at least confirm that it does or does not work? Thanks Ian – Ian H. 31 Sheerwater, Northampton NN3 5HU, Tel: 01604 513875 Preparing eBooks for Kindle and ePub to give the best reader experience.