155.247.166.220 downloads stalled

[Ichbin3]

And again ...

Code: Select all

16:32:10:WU01:FS00:Connecting to 155.247.166.220:8080
16:32:10:WU01:FS00:Downloading 3.75MiB
16:32:17:WU01:FS00:Download 5.01%
16:32:26:WU01:FS00:Download 6.68%
16:32:47:WU01:FS00:Download 8.34%
16:33:05:WU01:FS00:Download 11.68%

[parkut]

Stalled download, almost 3 hours now. Linux Client Version: 7.6.13, and Core Version: 0.0.11

Code: Select all

13:43:07:WU02:FS02:Connecting to assign1.foldingathome.org:80
13:43:07:WU02:FS02:Assigned to work server 155.247.166.220
13:43:07:WU02:FS02:Requesting new work unit for slot 02: READY gpu:0:GM206 [GeForce GTX 960] 2308 from 155.247.166.220
13:43:07:WU02:FS02:Connecting to 155.247.166.220:8080
13:43:08:WU02:FS02:Downloading 1.39MiB

[_r2w_ben]

On Windows, you can use TCPView to kill the stalled download. Find FAHClient.exe in the list, right click the connection with vav4.ocis.temple.edu as the Remote Address and then click Close Connection.

[Ichbin3]

I'm running linux.
Btw - the next stalled dl happend.
3 MB would have needed 1h hour to download.
I'm starting to hate that server.

[HaloJones]

Just applied a block on that IP to my firewall. In theory, the clients that are assigned to that server will fail to connect rather than connect but fail to download.

[scott@bjorn3d]

These servers are killing me doing work. Why don't they ever fix them?

[HaloJones]

server needs a frequently scheduled reboot. it's a simple cronjob

[Sparkly]

And another 3, so I am seriously considering blocking this server in my firewall permanently.

[Ichbin3]

I did now too

[rickoic]

Have been having small problem with hung downloads over the past week with this server, 1-3 times a day a reboot was required. Been living with it as it was a minor problem. Woke this morning and 3 of 4 pcs required a reboot. Total of 4 gpus hung. Just checked my pcs again and had 1 hung up so I rebooted. Got 220 server again and it hung with downloading 3.75gb. Sat for a few minutes with no progress, so I rebooted again. And again I was assigned to 220. Got to 1.8% downloaded this time and then it hung. Total of 6 reboots and on the final reboot it caused my other 3 gpus on the board to throw their wu's and redownload new ones. But at least the problem child got sent to another server and is working again.

[HaloJones]

How is this not being sorted despite all this noise????

[vvoelz]

Sorry for the ongoing problems with vav4. We recently put up more GPU WUs and the connection issue has worsened. I just did a hard reboot of this machine, which we will continue to do every day from now on. Its unclear whether this will actually do the trick, so if you observe ANY amelioration of problem, please post.

I have said before that we are trying to retire this server, and we still are. Hopefully our new hardware will installed in 1-2 months.

[rickoic]

Just had another failure to download from 220.

[vvoelz]

UGH - if we didn't have vital projects being served from vav4 I'd shut it down. (The server code does not let us easily migrate projects from one server to another). There must be some load balancing issues (collection server relays?) that are beyond our control, but perhaps we can set some better parameters. We'll continue to push on this.

[Sparkly]

UGH - if we didn't have vital projects being served from vav4 I'd shut it down. (The server code does not let us easily migrate projects from one server to another). There must be some load balancing issues (collection server relays?) that are beyond our control, but perhaps we can set some better parameters. We'll continue to push on this.

I would be surprised if this issue was a load-balancing thing, since it is more likely that it is a TCP packet loss/corruption/retransmission thing.

Could be as simple as a security setting in your firewall, so if you have older Cisco FWSM stuff on your boarders, you can try turning the sequence randomisation off.

https://community.cisco.com/t5/security ... n_and_SACK