Slow WU downloads & failed transfer [140.163.4.231]

[Aurum]

As WUs finish I'm not getting any new work, just as shown above.

[bollix47]

Currently, there are a number of server changes happening. As a result of this there are redirects and domain name servers (DNS) that may be stale in your various caches. How you clear those caches depend on the o/s or browser being used.

Other problems may be caused by security software needing to be updated with new exceptions ( you can test that by temporarily disabling that software ) or DNS servers not being updated in a timely manner. You may find that your problems subside as DNS servers get updated with new information but that depends on the ones you use. You could try changing your DNS servers to google ( 8.8.8.8,8.8.4.4 ). The only reason I suggest those is that I use them and am not having any problems getting work.

If you have trouble testing/changing any of the above let us know and we'll try to assist.

Here is a very recent snippet from one of my logs getting a new GPU WU:

Code: Select all

10:25:15:WU00:FS00:Connecting to 65.254.110.245:8080
10:25:15:WU00:FS00:Assigned to work server 140.163.4.231
10:25:15:WU00:FS00:Requesting new work unit for slot 00: RUNNING gpu:0:GP104 [GeForce GTX 1070] 6463 from 140.163.4.231
10:25:15:WU00:FS00:Connecting to 140.163.4.231:8080
10:25:16:WU00:FS00:Downloading 5.42MiB
10:25:17:WU00:FS00:Download complete
10:25:17:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR ...

[Aurum]

I've been using 8.8.8.8 & 8.8.4.4 on all my computers for a long time.
It doesn't seem to be a problem for my Win7 computers. Many but not all of my Linux Mint computers are affected.
I had previously disabled hibernation when I installed Linux but since a recent update it has returned to the shutdown menu. The same command no longer removes hibernation:

Code: Select all

sudo mv -v /etc/polkit-1/localauthority/50-local.d/com.ubuntu.enable-hibernate.pkla /

I've rebooted my router twice today and modem once.
I'm a Linux novice. Since these are headless DC rigs and not desktop PCs I also uninstalled several software packages Mint came with: Libre Office, Hex Chat, Pidgin Messenger, Simple Scan, Thunderbird Mail and Rhythmbox. The first time I did this I also uninstalled Graphics Tablet and had massive problems. Probably broke dependencies I do not know how to fix so I did a fresh Mint install and used the rule of thumb that if what I uninstalled had names that included Hexchat it was ok. If there was another name in the list of to be deleted packages then I won't touch that program.
Suggestions welcome.

[bollix47]

Try flushing your dns cache:

https://tecadmin.net/flush-dns-cache-ubuntu/

[OntosChalmer]

Wow, you're quite fortunate, having a connection able to download a GPU WU in a matter of a second. I tried flushing my DNS Cache. No improvement in speed or connectivity.

[toTOW]

1. My ISP uses a proxy by default.

This is often the root cause of any FAHClient connectivity issues ...

Other most common cause being security software needing exceptions to be added to leave FAH alone.

[OntosChalmer]

Correction. I used a proxy checking website, and there's... no proxy server?

[Aurum]

Try flushing your dns cache: https://tecadmin.net/flush-dns-cache-ubuntu/

That seems to work. Have to move a monitor from rig to rig since NoMachine works so slow. Did this:

Code: Select all

sudo /etc/init.d/dns-clean start
sudo service network-manager restart

So if flushing the DNS cache is good is disabling local caching even better???

[bollix47]

Local caching should speed up DNS searches, so no, I would not disable it. You could add the DNS flush to one of your startup files like rc.local so that any time you reboot you'll produce a fresh cache but in all likelihood you will not see this problem again any time soon and usually it sorts itself out like you've already seen on some of your systems.
Although we've seen a few reports of this problem considering there are thousands of folders the numbers having a problem are quite low. Personally, I had one problem from eight systems and a flush cured that one.

[Aurum]

I lost almost half my production for almost a day. Too bad there seems to be no way for F@H to send out a message recommending a DNS cache flush when they reconfigure.
Given that my rigs are dedicated headless computers running F@H 24x7 and hopefully rarely need rebooting I don't see why I even want to have a local DNS cache. After reading about DNS caches I see no benefit and only risk for a dedicated distributed computing rig. https://www.lifewire.com/what-is-a-dns-cache-817514
I'll be disabling my DNS caching. For Linux:

Code: Select all

sudo vi /etc/NetworkManager/NetworkManager.conf
Comment out:
#dns=dnsmasq
sudo service network-manager restart

[bruce]

This looks like a DNS failure associated with your ISP.

What you should be getting is 65.254.110.245 assign1.foldingathome.org

Are you able to open either http://65.254.110.245 and/or http://assign1.foldingathome.org? Both work for me.

By the way, we started talking about slow data transfer and failures associated with a specific WS and now you're talking about a DNS failure -- not knowing where to find an AS. These are two absolutely different problems.

[OntosChalmer]

Code: Select all

11:40:49:WU00:FS00:Connecting to 128.252.203.2:8080
11:46:19:WU02:FS00:Downloading 4.53MiB
11:46:31:WU02:FS00:Download 45.55%
11:46:43:WU02:FS00:Download 46.93%
11:46:55:WU02:FS00:Download 48.31%

Another WU assignment server. What confuses me is that it downloads so much % in 12 seconds, then slows down to a crawl.

[EXT64]

I am seeing the same. Typically a few times a day I have to manually kill the client and restart it to get a hung upload to reset. Typically within 1 try (sometimes 2) after a client restart it will upload instantly.

[Aurum]

Are you able to open either http://65.254.110.245 and/or http://assign1.foldingathome.org? Both work for me.

They both work for me.

By the way, we started talking about slow data transfer and failures associated with a specific WS and now you're talking about a DNS failure -- not knowing where to find an AS. These are two absolutely different problems.

Who cares?

[bruce]

Who cares?

People who follow the Forum Policies do.

Please pay attention to the intended topic of a thread and don't post off topic. Use the new topic button to start a new thread instead.

We try to be a helpful site for everyone, even if they're not as smart as you or if their native language is other than English. If folks are looking for a discussion of a DNS problem for the AS, they shouldn't have to open a topic entitled Slow downloads & failed transfer associated with a specific Work Server.