Server outrages

[gd_barnes]

Originally Posted by Flatlander

When the clients try to send results and the server is down it gives up after a few minutes and close the client. How do I extend that time?

As Max said, everything is back now. Yes, the outage was only 30 mins. this time so the UPS held its own and everything restarted automatically without the machine actually turning off. Now I just have to manually restart my other 11 machines. Ugh!

Chris, the clients should automatically keep trying once per minute forever until you hit Ctl-C.

[mdettweiler]

Originally Posted by Flatlander

When the clients try to send results and the server is down they give up after a few minutes and close the client. How do I extend that time?

Set op_connect=TRUE in do.bat. That will tell the client to keep retrying until the server comes back online. I'm actually not sure why that's set to FALSE by default...I suppose it could be useful if you want to have the computer fall back to some manual work (by stringing it up after the LLRnet client in a batch file) when the server goes down.

@Gary: that's how do.pl behaves. do.bat closes out after two failures (IIRC) with op_connect=FALSE, which is the default.

[mdettweiler]

Servers are down right now--probably another power outage. I am currently awaiting word from Gary to confirm this.

[mdettweiler]

Originally Posted by mdettweiler

Servers are down right now--probably another power outage. I am currently awaiting word from Gary to confirm this.

Just heard from Gary--it's not a power outage. Apparently, both the phone and internet went out simultaneously, which is rather odd since his internet is cable-based (so they're not on the same line or anything).

[mdettweiler]

Originally Posted by mdettweiler

Just heard from Gary--it's not a power outage. Apparently, both the phone and internet went out simultaneously, which is rather odd since his internet is cable-based (so they're not on the same line or anything).

Just heard something rather interesting from Gary...apparently the TV still works, which means that the cable line itself is functioning. Since power is on but telephone is not, that means that somehow, somewhere along the line, the cable internet is apparently dependent on the telephone connection! Strange but true.

Edit: Just realized something funny. Gary's cable company (Time Warner, which also services my area) has the usual cable phone service offering (which I don't believe Gary has) which they position as a direct competitor to standard phone company service. For the uninitiated, these are thinly disguised VoIP setups--they use the cable internet as a conduit for voice data. The funny thing is, though, if the cable internet goes out whenever the phone line goes out...that means that the cable phone service also goes out whenever the "real" phone line goes out. Go figure.

Edit 2: On second thought, scratch that...now that I think about it, I do remember Gary mentioning once that does use the cable phone service. In that case, it would not be at all surprising that the phone and internet would be out but not the power or TV. Rather, the cable line itself would seem to be functioning, but somewhere along the line the internet is not working. Thus, the phone wouldn't work either since its conduit is gone.

[gd_barnes]

Kind of being a Mr. drama queen there. lol The outage was < 30 minutes and wasn't even worth analyzing. Its entirely normal that the phone and internet are out but not the cable TV. I have Time Warner for all 3 and the phone and internet almost always go out together.

[MyDogBuster]

I've got just the opposite problem in Delaware (Comcast).

My cable TV goes up and down like a yoyo, but my phone and internet are ALWAYS on. Same company for all three, and when I call them they never have a problem BUT the cable comes back in about a minute after I hang up.
Go figure. Always about a minute. I can set my watch by it. Sounds like a re-boot on a balky server to me.

[Brucifer]

hmm.... appears that the servers are offline as I can't get a ping through on a new system I just brought online. :-( 6:40pm PST Tried both 3000 and 7000

[mdettweiler]

Originally Posted by Brucifer

hmm.... appears that the servers are offline as I can't get a ping through on a new system I just brought online. :-( 6:40pm PST Tried both 3000 and 7000

Strange...seems to be working now. And I have an open SSH session to one of Gary's other machines via the server (for GPU testing); that didn't get interrupted, so either there was an outage on your end or something else strange is keeping you from connecting.

Can you connect to the server now? If you can't get to the port 3000 and 7000 servers, can you load http://www.noprimeleftbehind.net/ in a browser?

Edit: Also, what is the IP address you're trying to connect from? This is just a hunch, but I figured I'd check just in case.

[Brucifer]

wasn't an ip address issue. Most likely if anything was a Slackware-64 related issue. The 64-suse box hooked up fine. But then I don't know cause the slack hooked up fine to other projects. ???? got me.

[mdettweiler]

Originally Posted by Brucifer

wasn't an ip address issue. Most likely if anything was a Slackware-64 related issue. The 64-suse box hooked up fine. But then I don't know cause the slack hooked up fine to other projects. ???? got me.

Very strange indeed. When you say that hte Slackware box connected fine to other projects, are you referring to other projects' LLRnet servers, or another type of application entirely? If the latter, I'd try recompiling LLRnet from source. (Once compiled, swap your new llrnet binary into the LLRnet2010 folder as previously installed.)