Mac OS: WU not pausing correctly

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0
Topic 189760

I don't know whether the following problem is with the E@H application, the BOINC 4.43 GUI, or Predictor. I've been impressed with the involvement of the UWM staff in figuring stuff out, so I'm starting here.

Symptom:
Observed in the BOINC Manager Work tab:
Project, Application, CPU time, Progress, To Completion, Status
E@H, einstein 4.82, 1:16 (and advancing), 10.7% (and advancing), 10:38:08 (and declining) Paused
P@H, mfoldB125 4.29, ---, 0.00%, ---, Running

In the Messages tab (where for some reason the "Copy Messages" buttons aren't doing anything), the E@H WU was supposedly paused 30 minutes ago (left in memory), and the P@H WU was resumed at that time. No error messages.

Activity Monitor shows:
228 einstein_4.82_po alanng 58.60 3 2.94 MB 44.94 MB
200 BOINCManager alanng 23.80 1 9.73 MB 111.92 MB
246 Activity Monitor alanng 6.10 2 26.46 MB 108.87 MB
206 boinc alanng 0.90 1 1.79 MB 27.56 MB
230 mfoldB125_4.29_p alanng 0.00 1 7.64 MB 92.64 MB
Most of the CPU load (more than half) is Nice, 5-10% System, rest is User.

Machine info:
http://einsteinathome.org/host/205881

When I exit BOINC and restart it, the E@H WU is paused correctly and that Predictor WU is running normally. I'll report back if this happens again.

Are you a musician? Join the Musicians team.
Meet the Musicians

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4266
Credit: 244924143
RAC: 16679

Mac OS: WU not pausing correctly

Hm, I thougth we had this fixed by now (see the MacOS Beta test thread in Cafe Einstein). I'm currently runnning the 4.82 on 10.4.2 without problems, but not with Predictor. I'll give it a try and see what happens.

Oh, btw.: I think the AEI staff is as important for the project as the UWM :-)

BM

BM

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

Yes, sorry, I should have

Yes, sorry, I should have said "E@H staff" instead of "UWM staff." Here's another instance of a similar problem. This time on a different machine, but again while attached to Predictor. Maybe this is indeed a Predictor problem?

Symptom:

Summary: BOINC is frozen.

BOINCManager Work tab shows:
E@H: App einstein 4.82, CPU Time ---, Progress 0.00%, To completion ---, Status Paused

P@H: App mFoldB125 4.29, CPU Time 5:33:53, Progress 76.13%, To completion 1:44:39, Status Running

Neither of the above stats are progressing (watched for several minutes). BOINCManager status is "Run based on preferences."

Activity Monitor shows system is idle ~90%. Relevant threads are:
401 BOINCManager alan 6.50 1 12.12 MB 123.10 MB
402 boinc alan 0.50 1 3.17 MB 27.56 MB
495 einstein_4.82_po alan 0.00 3 5.44 MB 44.94 MB
486 mfoldB125_4.29_p alan 0.00 1 41.14 MB 92.64 MB

Then, when I exit and restart BOINC Manager, the P@H WU resumes normally, with the E@H paused normally.

Machine info:
http://einsteinathome.org/host/335019

Using BOINC Standard GUI 4.43.

If you think we should contact P@H staff about this, go ahead or ask me to post in a P@H forum a link to this forum thread.

Are you a musician? Join the Musicians team.
Meet the Musicians

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

Another case of the same

Another case of the same problem (E@H WU not running while Predictor WU is supposedly paused):

Predictor WU: mfoldB125 4.29 stats advancing at 00:02:56+, 11%+, "Paused"

Einstein WU: einstein 4.82 stats frozen at ---, 0.00%, ---, "Running"

Activity Monitor reports:
1049 mfoldB125_4.29_p alan 38.00 1 41.14 MB 92.64 MB
1050 einstein_4.82_po alan 36.80 3 6.50 MB 44.94 MB
983 BOINCManager alan 7.10 1 12.59 MB 123.65 MB
984 boinc alan 0.60 1 3.36 MB 27.56 MB

Almost all of the CPU load is Nice, 5-10% System, and the rest User.

Machine info:
http://einsteinathome.org/host/335019
(Mac Mini running BOINC Standard GUI 4.43, set to "Run based on preferences," attached only to E@H and P@H).

Are you a musician? Join the Musicians team.
Meet the Musicians

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

Observed another case today

Observed another case today of the exact same problem as below (8/29/05 0:48:31), but this time on a Powerbook:
http://einsteinathome.org/host/205881

Are you a musician? Join the Musicians team.
Meet the Musicians

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4266
Credit: 244924143
RAC: 16679

I am running Predictor and

I am running Predictor and Einstein on my PowerBook on 10.4.2 right now and haven't seen the problem yet.

I'll continue to look at this. If you get annoyed, try to set "leave applications in memory" in your preferences to "no".

BM

BM

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

A new manifestation of

A new manifestation of similar issue. This time the context is:
Mac mini: http://einsteinathome.org/host/335019

Attached to: E@H 30, P@H 15, Seti@H 25
(note change from previous posts is that I re-attached to Seti@H as well. I never observed this problem before when this machine was attached to Seti@H.)
Running BOINC Standard GUI 4.43
Set to "Run based on preferences," "Leave in memory"

Work tab shows:
einstein 4.82, with advancing stats for CPU Time, Progress, and To completion, Status "Running"
seti 4.18, all stats at ---/0%, Status "Paused,"
predictor mFoldB125 4.29, 00:54:25, 40.04%, 01:21:30, Status "Paused,"

That all looks ok until I look at the Messages tab and see the following preceding sequence of events (gee, it would be nice if the "Copy messages" buttons worked):
...
Einstein 14:10:12 Pausing result (left in memory)
Seti 14:10:12 Starting result
15:10:12 schedule_cpus: time 3600.065725
ProteinPredictor 15:10:12 Resuming result
Seti 15:10:12 Pausing result (left in memory)
...

Thus apparently einstein 4.82 paused at 14:10:12, and when seti 4.18 ran for one hour, nothing happened. Seems to me this problem may be relevant to this thread?

Activity monitor shows currently that the seti and predictor apps are properly, completely idle at the moment, while einstein is running properly.

Are you a musician? Join the Musicians team.
Meet the Musicians

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

Yet a different flavor of

Yet a different flavor of same bug...?

Powerbook: http://einsteinathome.org/host/205881
Attached to: E@H 30, P@H 15
Running BOINC Standard GUI 4.43, "Run based on prefs," Leave in memory"

In Work tab I observed P@H and E@H WUs running simultaneously. P@H status was "Paused" but its stats were advancing. E@H status was "Running" and its stats were also advancing.

Possibly relevant background history was that Messages showed that the previous E@H WU had finished properly but without the "finished file."

Reboot caused the P@H and E@H WUs to resume correct behavior where they had each left off before reboot.

Are you a musician? Join the Musicians team.
Meet the Musicians

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

New variation on problem

Message 15913 in response to message 15911

New variation on problem described below (Posted 1 Sep 2005 21:45:47 UTC)

Again on the same Mac mini: http://einsteinathome.org/host/335019

Attached to: E@H 30, P@H 15 (note: no longer attached to S@H)
Running BOINC Standard GUI 4.43
Set to "Run based on preferences," "Leave in memory"

Work tab shows:
einstein 4.82, all stats at ---/0%, Status "Paused,"
predictor mFoldB125 4.29, with advancing stats for CPU Time, Progress, and To completion, Status "Running"

As before, the Messages tab shows that apparently einstein 4.82 ran for one hour just previously, before paused in order to swap to the P@H WU, but the E@H WU did not change its stats.

This time, however, Activity Monitor shows that all BOINC apps are completely idle (0.00%). BOINCManager is taking the usual ~5% and boinc is taking the usual <1%.

Exiting and restarting BOINCManager then shows E@H WU still at zero stats, while P@H resumes normally (confirmed both with advancing Work tab stats and also in Activity Monitor).

Are you a musician? Join the Musicians team.
Meet the Musicians

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

Encouraging news: I've

Encouraging news:

I've continued running Predictor/Einstein on the Powerbook mentioned below, but in the past week since I did a "Reset Project" on E@H these problems have not reoccurred.

Also: on the Mac mini I switched to running Einstein/SETI instead of Einstein/Predictor for about a week and the problems did not happen then, either. However, I did execute a "Reset Project" on E@H on that machine first.

Perhaps "Reset project" is the advisable first-aid step when a user observes a problem like those described in this thread.

Are you a musician? Join the Musicians team.
Meet the Musicians

Alan Ng
Alan Ng
Joined: 19 Mar 05
Posts: 34
Credit: 75464274
RAC: 0

Bad news: Now I've

Bad news:

Now I've observed this problem when connected to E@H and Seti@H only. This eliminates Predictor from being the source of the problem.

Machine info (Mac mini):
http://einsteinathome.org/host/335019

Work tab says:
einstein 4.82; w1_0954...; 00:01:25 (and climbing); 6.28% (and climbing); 00:22:08 (and changing); "Paused"
SETI@home setiathome 4.18 (altivec optimized app); 06nov03aa....; ---; 0.00%; ---; "Running"

Activity Monitor says:
setiathome ~43% CPU
einstein ~42% CPU
BOINCManager 5% CPU
boinc ~0.5% CPU

Total load is almost all "Nice."

I observed this as soon as I attached to SETI and got a WU just a bit ago. Previously this machine has been running solely E@H for a couple weeks, quite peacefully.

When I exit BOINC, all BOINC-related threads are properly killed. When I restart BOINC, correct behavior resumes (the E@H is truly paused and the S@H WU is progressing normally, including in the BOINCManager Work stats).

Are you a musician? Join the Musicians team.
Meet the Musicians

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.