Sporadic issues with 171.67.108.xx servers

Moderators: Site Moderators, FAHC Science Team

foldy
Posts: 2061
Joined: Sat Dec 01, 2012 3:43 pm
Hardware configuration: Folding@Home Client 7.6.13 (1 GPU slots)
Windows 7 64bit
Intel Core i5 2500k@4Ghz
Nvidia gtx 1080ti driver 441

Re: Sporadic issues with 171.67.108.xx servers

Post by foldy »

Any chance to try a different LAN switch? But it still may be something totally different.
ComputerGenie
Posts: 236
Joined: Mon Dec 12, 2016 4:06 am

Re: Sporadic issues with 171.67.108.xx servers

Post by ComputerGenie »

foldy wrote:Any chance to try a different LAN switch? But it still may be something totally different.
For entirely unrelated reasons (the 8-port that was there was needed elsewhere), the 4-port that they share was just replaced on Saturday (and my issue started on Wednesday). And I'm nearly certain that there are 0 issues with the 24-port that the Win 7 box is on.
I'm at wits end over this, mostly due to what all else is connected and working in the totality of the networking. :(
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Sporadic issues with 171.67.108.xx servers

Post by bruce »

Does your router (or Proxy) use a smart protocol that drops "unnecessary" packets? I've seen cases where that caused errors.

I'm particularly interested in any one thing that's between each of your machines and the internet so the idea that it might be a flaky switch was a good one.
_r2w_ben
Posts: 285
Joined: Wed Apr 23, 2008 3:11 pm

Re: Sporadic issues with 171.67.108.xx servers

Post by _r2w_ben »

Do you by any chance have multiple ISPs with load balancing between the connections?
ComputerGenie
Posts: 236
Joined: Mon Dec 12, 2016 4:06 am

Re: Sporadic issues with 171.67.108.xx servers

Post by ComputerGenie »

bruce wrote:Does your router (or Proxy) use a smart protocol that drops "unnecessary" packets? I've seen cases where that caused errors.

I'm particularly interested in any one thing that's between each of your machines and the internet so the idea that it might be a flaky switch was a good one.
The routers are just C7s (issue remains with "DoS Protection" on or off; as far as I know, that's the only such filter of that type) and we use base-model TP-Link switches (which show all "bad packets" rows to read 0).
_r2w_ben wrote:Do you by any chance have multiple ISPs with load balancing between the connections?
Sadly, 1 ISP (and 1 connection per) is all we can get here.
foldy
Posts: 2061
Joined: Sat Dec 01, 2012 3:43 pm
Hardware configuration: Folding@Home Client 7.6.13 (1 GPU slots)
Windows 7 64bit
Intel Core i5 2500k@4Ghz
Nvidia gtx 1080ti driver 441

Re: Sporadic issues with 171.67.108.xx servers

Post by foldy »

If we cannot find out the reason for the stuck download problem, a workaround could be a script which checks every 1 min if a gpu slot is idle then waits for 20 min if it gets running again (max download time is 15min I think until server closes connection). If it is not running again then the FahClient is restarted.

I'm not sure about the server timeout of a download, is it 15min? Because if we restart too early then a good big download may never finish in our timeout.

A similar workaround could be implemented in the FahClient itself except that it would only cancel the download and restart the slot.
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Sporadic issues with 171.67.108.xx servers

Post by bruce »

foldy wrote:I'm not sure about the server timeout of a download, is it 15min?
The default server timeout is, in fact, 15 minutes. If an upload or download cannot be completed by the timeout, the connection is dropped.

In the example posted on the previous page, we see

Code: Select all

11:36:24:WU01:FS01:Connecting to 171.67.108.102:8080
11:36:34:WU01:FS01:Downloading 7.06MiB
11:36:40:WU01:FS01:Download 94.77%
11:36:40:WU01:FS01:Download complete
This took 6 seconds on his ISP, so it's not a factor. Ideally, the protocol recognizes that the download was completed ("Download complete") and the connection closes naturally. Otherwise, the connection remains "active" for another 14m54s before being forcefully dropped.

There are some cases where the server timeout is adjusted (increased). The goal is to down-/up-load the data package over especially slow donor connections without unnecessary interruptions.
ComputerGenie
Posts: 236
Joined: Mon Dec 12, 2016 4:06 am

Re: Sporadic issues with 171.67.108.xx servers

Post by ComputerGenie »

Well, as inexplicably as it started, it stopped. Not a single problem in the last 48 hours.
Thanks to all that tried to help. Good folding to all...
ComputerGenie
Posts: 236
Joined: Mon Dec 12, 2016 4:06 am

Re: Sporadic issues with 171.67.108.xx servers

Post by ComputerGenie »

foldy wrote:If we cannot find out the reason for the stuck download problem, a workaround could be a script which checks every 1 min if a gpu slot is idle then waits for 20 min if it gets running again (max download time is 15min I think until server closes connection). If it is not running again then the FahClient is restarted...
Well, it looks like I'm going to have to get off my lazy butt and write that, because I'm having it all over again. :cry:
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Sporadic issues with 171.67.108.xx servers

Post by bruce »

It would be helpful to identify SPECIFIC servers that are having this problem. I suspect that the problems are not shared equally across the various 171.67.108.xx servers. f possible, specific error message(s) associated with each server would help, too.

From the previous page, this is lacking that sort of information
ComputerGenie wrote:
rwh202 wrote:That log shows a failed download from 171.67.108.157:8080 and a successful one from 171.67.108.102:8080

Have you ever had success from .157 or a failure from another server that has sometimes worked e.g. .102...
It has happened with more than one (ironically, .102 is usually the most afflicted; however, that could just be because it's also the one most designated).
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Sporadic issues with 171.67.108.xx servers

Post by bruce »

ComputerGenie wrote:Is there a possible chance that it's a time sync issue (either in the software or on my end)?
I'm looking at the most recent WU...
on the "Status" tab, it says "Assigned: 2017-06-12T19:02:46Z"
in the log is: "19:02:22:WU00:FS00:0x21:Completed 187500 out of 6250000 steps (3%)"
Meaning that 24 seconds before status claims it was assigned, it was 3% done. :shock:
The "Assigned" time (and eventually, the Upload-Complete time) are based on the Server's clock. (That way adjusting your local clock does not alter the deadlines or the bonus calculation.) The timestamps in your log are based on your local clock.

Apparently your local clock was not aligned with the server clock, but that shouldn't cause any significant problems..
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Sporadic issues with 171.67.108.xx servers

Post by bruce »

bruce wrote:1) please reset verbosity to the default value.
2) The key piece of information in that log is GPUs: 0. Why do you need to restart?
Did I miss your answers somewhere?
ComputerGenie
Posts: 236
Joined: Mon Dec 12, 2016 4:06 am

Re: Sporadic issues with 171.67.108.xx servers

Post by ComputerGenie »

bruce wrote:
bruce wrote:1) please reset verbosity to the default value.
2) The key piece of information in that log is GPUs: 0. Why do you need to restart?
Did I miss your answers somewhere?
That 0 was do to a clearing of files. Basically, in FAHClient's mind it was a 1st run. Successive boots showed correct # of GPUs.
The why reboot and why dump files was there not much left to try and I was willing to try anything.

Seems "Cancer" is the answer for me; Parkinson's, Alzheimer's, and "Any" all lead to the no download issue (and Huntington's was "hit and miss" as to if it would work right or not). I'm not sure why it is that way, but after this much time messing with it and this many hours of no science being done (vs the Parkinson's that I actually want and the reason I got started folding), any science filed is better than nothing.
Post Reply