Bugzilla – Full Text Bug Listing
| Summary: | Long pause on first connection to orion.eclipse.org | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [ECD] Orion | Reporter: | John Arthorne <john.arthorne> | ||||
| Component: | Releng | Assignee: | Project Inbox <orion.releng-inbox> | ||||
| Status: | RESOLVED WORKSFORME | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | P3 | CC: | jainnayna, Mike_Wilson, simon_kaegi, webmaster | ||||
| Version: | unspecified | ||||||
| Target Milestone: | --- | ||||||
| Hardware: | PC | ||||||
| OS: | Windows 7 | ||||||
| Whiteboard: | |||||||
| Attachments: |
|
||||||
|
Description
John Arthorne
CCing webmaster in case this is a server configuration problem. Nobody has seen this on localhost servers, but I don't know if we have samples of other remote servers to help us decide if it is specific to eclipse.org or just a local vs remote server problem. McQ reported that when this happened, he switched to another browser tab and immediately the request worked (while the original tab was still hung). It's like the first connection didn't receive an ack so it hangs... Created attachment 198151 [details]
packet dump
Someone is definitely talking too much, and the other ain't listening. Time for counseling.
Denis, can you help decipher this for us networking dummies? Is this a view from the client side? Is it that the client sends HTTP GET / and the server never responds? This is on my client computer. The 3-way handshake is established (frames 1-3) and in frame 4 my computer requests the home page. The server then sends a 3-way handshake ACK packet (again) and client advises it's a dupe. Server repeats this twice again, and then gives up. Poor client is left in the dark, desperately seeking the attention from the server, which has given up. This is likely a network glitch. When I originally copied the orion.eclipse image to orionhub, I forgot to set a unique MAC address for the virtual network card. That had both servers responding to the same packets. I get the feeling that I may need to restart both the host guest servers and even the host server to clear this up. I've been trying to get a packet dump from the first router after the server to see what's happening there, but every time I try to access the Orion vservers I don't get the symptomatic endless response. I've had a look at this, and it seems orion.eclipse and orionhub are confused. My next step would be to restart the host server to clear out any remaining artifacts of the initial arp address conflict that may be left behind. Unfortunately, the same host server is also hosting our bittorrent instance, which is needed for the Indigo launch. Another option would be for me to provision you a brand new, blank server image, and you can install orion it it. If you want to restart the server today before Indigo goes live, feel free. Otherwise I guess we'll need to go the new server route to avoid disrupting the torrents during the post-release rush. Had this problem occur again today during a demo. Is there something more we can do to help figure out what's going on? *** Bug 351450 has been marked as a duplicate of this bug. *** This problem was resolved through no change in Orion. |