Cannot connect to get jobs

[aardWolf]

My employer blocks tons of outgoing ports. This was never a problem with the older versions of the F@H client. I just updated the client to 7.4.4, and now it's a problem. It's timing out attempting to connect to 171.67.108.200:8080. This means it never gets new work units.

I've been looking for a place to change that connection. Is there anywhere I can connect on port 80?

[Joe_H]

The folding client should also be trying on port 80. Could you post your log showing the attempts to download work as well as the beginning section showing your system information and configuration settings?

[aardWolf]

This is all it is showing... Our network is set up in such a way that things don't time out for quite a while (20-30 minutes). Nothing has happened for the past 17 minutes in the logs (yes, I clicked Refresh).

Code: Select all

*********************** Log Started 2014-04-01T15:44:36Z ***********************
15:44:36:************************* Folding@home Client *************************
15:44:36:    Website: http://folding.stanford.edu/
15:44:36:  Copyright: (c) 2009-2014 Stanford University
15:44:36:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
15:44:36:       Args: --child --lifeline 90450 --respawn
15:44:36:     Config: /Library/Application Support/FAHClient/config.xml
15:44:36:******************************** Build ********************************
15:44:36:    Version: 7.4.4
15:44:36:       Date: Mar 4 2014
15:44:36:       Time: 20:27:54
15:44:36:    SVN Rev: 4130
15:44:36:     Branch: fah/trunk/client
15:44:36:   Compiler: GNU 4.2.1 (Apple Inc. build 5666) (dot 3)
15:44:36:    Options: -std=gnu++98 -O3 -funroll-loops -mfpmath=sse -ffast-math
15:44:36:             -fno-unsafe-math-optimizations -msse3 -arch x86_64
15:44:36:             -mmacosx-version-min=10.6
15:44:36:   Platform: darwin 10.8.0
15:44:36:       Bits: 64
15:44:36:       Mode: Release
15:44:36:******************************* System ********************************
15:44:36:        CPU: Intel(R) Xeon(R) CPU E5645 @ 2.40GHz
15:44:36:     CPU ID: GenuineIntel Family 6 Model 44 Stepping 2
15:44:36:       CPUs: 24
15:44:36:     Memory: 12.00GiB
15:44:36:Free Memory: 967.14MiB
15:44:36:    Threads: POSIX_THREADS
15:44:36: OS Version: 10.9
15:44:36:Has Battery: false
15:44:36: On Battery: false
15:44:36: UTC Offset: -5
15:44:36:        PID: 90490
15:44:36:        CWD: /Library/Application Support/FAHClient
15:44:36:         OS: Darwin 13.1.0 x86_64
15:44:36:    OS Arch: AMD64
15:44:36:       GPUs: 1
15:44:36:      GPU 0: ATI:4 Juniper [Radeon HD 5700 Series]
15:44:36:       CUDA: Not detected
15:44:36:***********************************************************************
15:44:36:<config>
15:44:36:  <!-- Folding Slot Configuration -->
15:44:36:  <cause v='ALZHEIMERS'/>
15:44:36:
15:44:36:  <!-- Network -->
15:44:36:  <proxy v=':8080'/>
15:44:36:
15:44:36:  <!-- Slot Control -->
15:44:36:  <power v='MEDIUM'/>
15:44:36:
15:44:36:  <!-- User Information -->
15:44:36:  <passkey v='********************************'/>
15:44:36:  <team v='3446'/>
15:44:36:  <user v='LawsonCulver'/>
15:44:36:
15:44:36:  <!-- Folding Slots -->
15:44:36:  <slot id='0' type='CPU'/>
15:44:36:</config>
15:44:36:Trying to access database...
15:44:37:Successfully acquired database lock
15:44:37:Enabled folding slot 00: READY cpu:23
15:44:37:WU00:FS00:Connecting to 171.67.108.200:8080
15:44:42:16:127.0.0.1:New Web connection

Mod edit: Added Code tags to log file

[bruce]

Welcome to foldingforum.org, aardWolf.

You bring up two different issues.

1) FAH is designed to communicate through either port 8080 or port 80. Many firewalls block port 8080. FAH's first choice is to use port 8080 but it's designed to revert to port 80 if port 8080 is blocked. A connection to 171.67.108.200 is not required and in most circumstances, the client will revert to a secondary assignment server at 171.64.65.121 which serves the same function through port 80. How long did you wait after the 15:44;42 UTC timestamp?

2) Your employer has the right to determine how his equipment is used. If you're using a computer that he owns, you must get his permission to run FAH on it. See the EULA. That may mean that you need to fold only at home. Stanford University is not interested in getting you fired for violating company policy, so do get (written) permission before using company resources for FAH.

[aardWolf]

1. It has now been 32 minutes since the connection attempt with no change in the logs.

2. My employer blocks ports wholesale to keep us "safe", but my Director is perfectly OK with us using our spare cycles for F@H. I told her about it when I first installed it quite a few months ago. We work for the department of Electronic Research Administration at the University of Tennessee Health Science Center though... so they're typically fine with us contributing to any kind of research.

[aardWolf]

As a side note... I had already let it sit for around an hour with nothing downloaded before uninstalling/reinstalling the same version of the client. So I did wait long enough for it to have timed out before posting.

[aardWolf]

After a reboot, it's at least timing out. Is there a reason that it is getting connection refused from 171.64.65.121:80?

Code: Select all

*********************** Log Started 2014-04-01T16:57:50Z ***********************
16:57:50:************************* Folding@home Client *************************
16:57:50:    Website: http://folding.stanford.edu/
16:57:50:  Copyright: (c) 2009-2014 Stanford University
16:57:50:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
16:57:50:       Args: --child --lifeline 125 --respawn
16:57:50:     Config: /Library/Application Support/FAHClient/config.xml
16:57:50:******************************** Build ********************************
16:57:50:    Version: 7.4.4
16:57:50:       Date: Mar 4 2014
16:57:50:       Time: 20:27:54
16:57:50:    SVN Rev: 4130
16:57:50:     Branch: fah/trunk/client
16:57:50:   Compiler: GNU 4.2.1 (Apple Inc. build 5666) (dot 3)
16:57:50:    Options: -std=gnu++98 -O3 -funroll-loops -mfpmath=sse -ffast-math
16:57:50:             -fno-unsafe-math-optimizations -msse3 -arch x86_64
16:57:50:             -mmacosx-version-min=10.6
16:57:50:   Platform: darwin 10.8.0
16:57:50:       Bits: 64
16:57:50:       Mode: Release
16:57:50:******************************* System ********************************
16:57:50:        CPU: Intel(R) Xeon(R) CPU E5645 @ 2.40GHz
16:57:50:     CPU ID: GenuineIntel Family 6 Model 44 Stepping 2
16:57:50:       CPUs: 24
16:57:50:     Memory: 12.00GiB
16:57:50:Free Memory: 10.08GiB
16:57:50:    Threads: POSIX_THREADS
16:57:50: OS Version: 10.9
16:57:50:Has Battery: false
16:57:50: On Battery: false
16:57:50: UTC Offset: -5
16:57:50:        PID: 253
16:57:50:        CWD: /Library/Application Support/FAHClient
16:57:50:         OS: Darwin 13.1.0 x86_64
16:57:50:    OS Arch: AMD64
16:57:50:       GPUs: 1
16:57:50:      GPU 0: ATI:4 Juniper [Radeon HD 5700 Series]
16:57:50:       CUDA: Not detected
16:57:50:***********************************************************************
16:57:50:<config>
16:57:50:  <!-- Folding Slot Configuration -->
16:57:50:  <cause v='ALZHEIMERS'/>
16:57:50:
16:57:50:  <!-- Network -->
16:57:50:  <proxy v=':8080'/>
16:57:50:
16:57:50:  <!-- Slot Control -->
16:57:50:  <power v='MEDIUM'/>
16:57:50:
16:57:50:  <!-- User Information -->
16:57:50:  <passkey v='********************************'/>
16:57:50:  <team v='3446'/>
16:57:50:  <user v='LawsonCulver'/>
16:57:50:
16:57:50:  <!-- Folding Slots -->
16:57:50:  <slot id='0' type='CPU'/>
16:57:50:</config>
16:57:50:Trying to access database...
16:57:51:Successfully acquired database lock
16:57:51:Enabled folding slot 00: READY cpu:23
16:57:52:WU00:FS00:Connecting to 171.67.108.200:8080
16:58:38:17:127.0.0.1:New Web connection
16:59:09:WARNING:WU00:FS00:Failed to get assignment from '171.67.108.200:8080': Failed to connect to 171.67.108.200:8080: Operation timed out
16:59:09:WU00:FS00:Connecting to 171.64.65.121:80
16:59:09:WARNING:WU00:FS00:Failed to get assignment from '171.64.65.121:80': Failed to connect to 171.64.65.121:80: Connection refused
16:59:09:ERROR:WU00:FS00:Exception: Could not get an assignment
16:59:09:WU00:FS00:Connecting to 171.67.108.200:8080
17:00:25:WARNING:WU00:FS00:Failed to get assignment from '171.67.108.200:8080': Failed to connect to 171.67.108.200:8080: Operation timed out
17:00:25:WU00:FS00:Connecting to 171.64.65.121:80
17:00:25:WARNING:WU00:FS00:Failed to get assignment from '171.64.65.121:80': Failed to connect to 171.64.65.121:80: Connection refused
17:00:25:ERROR:WU00:FS00:Exception: Could not get an assignment
17:00:25:WU00:FS00:Connecting to 171.67.108.200:8080

Here's a traceroute:

Code: Select all

$ traceroute 171.64.65.121
traceroute to 171.64.65.121 (171.64.65.121), 64 hops max, 52 byte packets
 1  172.21.64.3 (172.21.64.3)  52.044 ms  0.497 ms  0.430 ms
 2  * * *
 3  * * *

[aardWolf]

I uninstalled 7.4.4, then reinstalled 7.3.6. I'm getting "Empty work server assignment" when it connects to assign4.stanford.edu:80.

[Joe_H]

I looked over your log a bit more closely. As you are running a high end OS X system you have a higher than usual core count. At the moment a few work servers are down that had OS X work available for high core counts. I would suggest setting the thread (core) count down to 12, there are available projects at that setting. For the other thread where this came up, see - viewtopic.php?f=18&t=26113.

P.S. At the slider setting of Medium, your machine is set to use 23 threads which is a large prime. I would suggest manually setting that to a different number when the large WU servers for OS X come back, 16 or 20 might be best.

[aardWolf]

I'm back on 7.4.4. At a thread count of 12 (for Alzheimer's research), it's still getting an empty work server assignment.

I dropped it down to 6 threads, and then 1 thread, and I'm still getting an empty assignment.

[billford]

At the moment a few work servers are down that had OS X work available for high core counts.

25% of all the GPU work servers are down

[aardWolf]

The OS X client doesn't support GPU anyway, correct?

[billford]

The OS X client doesn't support GPU anyway, correct?

Oops- I got confused as to which topic I was reading

[bruce]

For those who are allowed to use port 8080, there's redundancy since either Assignment Server 171.67.108.200 or 171.64.65.121 can search for a work server that has WUs for you. When you're restricted to port 80, you do not have that extra redundancy but having either AS out of service is pretty rare.

Once you successfully connect to an AS, it has to find a Work Server that has WUs that your machine can process. At that level, redundancy is provided by having multiple WS with WUs for (in your case) CPUs on OS-X. If all such Work Servers are off-line or all such projects are out-of-WUs to assign, you'll get the message "Empty work server assignment." That condition is generally solved quickly either by (1) bringing one or more servers back on-line, (2) by WUs currently being process being returned, thereby creating a new next-generation WU, or (3) by bringing new projects for OS-X on-line.

Item (1) may be prompt or may take more time, depending on why the server is off-line.
Item (2) is a continual process requiring no action by the Pande Group and is especially quick for projects which are shared between Windows, Linux, and OS-X.
Item (3) generally takes quite a bit of the scientist's time so it's rarely I would call prompt but it's also an ongoing process.

Is your system configured for client-type=advanced or are you using the default setting?

If you see "Empty work server assignment." frequently, report the time and duration of the problem.

Although this can be done later, if you're interested, you may choose to capture the information from serverstat, both before and after the next update following when you see the problem. Ignore all GPU servers. Categorize the remaining servers into four categories: {Full+Accepting vs. Not-Full and/or Not-Accepting} and {One or more X is in the right-most white column vs. no X appears there}. That information may help solve future problems. Additional analysis will also be needed, but that's a good start. Of the Full+Accepting+X servers, have large or small numbers in the WU_Avail column?

[aardWolf]

Is your system configured for client-type=advanced or are you using the default setting?

Since the initial install, I've just accepted the default settings. The Threads was set to -1, which let the system choose how many to use. I also chose Alzheimer's research specifically.

I noticed that it didn't get work units even with the threads set to 1 and not specifying the type of research. Maybe it's out of OS X units entirely...

Also, I've never seen the server status thing before... so it's a bit over my head.
http://fah-web.stanford.edu/pybeta/serverstat.html