Community
Participate
Working Groups
At times when I'm editing on org.eclipse.orion, I'll get save timeouts. Just now I got about 5 timeouts in a row and went to the navigator to make sure the server was alive. It was. We need to either increase the timeout or use the progress service. I think the latter would be better so the user knows what's going on. Right now, if it times out, the dirty indicator stays up but otherwise there's no way to know that it gave up without going to the console.
To narrow down the root cause of this, can you try to use orion.eclipse.org:8080 instead? The difference is that port 80 goes through the Apache reverse proxy and port 8080 does not.
I'll work from 8080 today and report back.
For file save, we're not likely to get a big benefit using the progress service. We need to maintain the server connection for the duration so the server can suck up the bytes. It's not a request we can accept and then process without maintaining a connection to the client. So, using the progress message in the UI would ensure a message appears to the user, but it wouldn't prevent timeouts from happening.
makes sense. so we need to: - understand why it's happening - possibly increase timeout - definitely report the error
I worked today using port 8080. For what it's worth, I didn't experience any timeouts on save today. Next week I'll try to run some pages on port 8080 and some without, and see if I notice a correlation.
I'm working on a very sketchy internet connection while travelling, and I find that most of orion.eclipse.org is working/can recover from connection glitches. But the save timeout is what's killing me. If there's a hiccup anywhere I reach the timeout. I hacked the timeout in the source and tried to edit from a self hosted site, but that site runs slower so it's even less usable. Oddly, there are certain files that always hit the timeout, and other files that I can save easily, and it's not a matter of the ones that work being smaller, in fact they are twice as big. I'm assuming this is coincidental. I think there are three things to do here: - we need to report the timeout error in the UI - using the progress service would let the user see that something is happening. I think a byte count could be useful, because then I'd know if things are just slow or if there are intermittent pauses. Of course we would want to make sure the progress reporting wasn't too noisy when one is on a good connection. - ideally I could tweak the timeout. It doesn't have to be a mainstream thing in the UI, but having a setting that I could hack would help. (Incidentally, other bandwidth soaking operations, like cloning all of the orion client repo, have worked. The connection is challenging/slow, but it's only the editor save that makes it completely unusable.)
(In reply to comment #6) > (Incidentally, other bandwidth soaking operations, like cloning all of the > orion client repo, have worked. The connection is challenging/slow, but it's > only the editor save that makes it completely unusable.) Git operations require only very small client/server bandwidth. All the bandwidth in these cases is between Git server and orion workspace server. GET and PUT on file contents are probably the only operations we have that ever transfer more than 1KB in a single request/response.
(In reply to comment #7) > Git operations require only very small client/server bandwidth. All the > bandwidth in these cases is between Git server and orion workspace server. GET > and PUT on file contents are probably the only operations we have that ever > transfer more than 1KB in a single request/response. I wasn't very clear here. I had to switch and run Orion on localhost. When I cloned the orion client git repo to my localhost workspace server (downloading the repo) the transfer was solid. (Haven't tried to do any fetching/merging though)
*** Bug 365939 has been marked as a duplicate of this bug. ***
This bug made it impossible for me to work home using orion.eclipse.org
Closing as part of a mass clean up of inactive bugs. Please reopen if this problem still occurs or is relevant to you. For more details see: https://dev.eclipse.org/mhonarc/lists/orion-dev/msg03444.html
