Java error

[DanielMahu]

Hello!

I've just installed the java client and I got this error -- probably at the moment it tries to send results.


2004.01.28 18:27:42.350
P0, Gen: 44, Top: F<f>0.9999999999999969</f><p0>0.0</p0><p1>0.0</p1><p2>0.6785714285714286</p2><g>04c16FXA8gs0X802LIOu4_YK5Ds9200UBoH[0JtFhToM1X^acgF9FGJNtuIUW]frl3eaQHIat7F52o1UeX26TK]FX1_a`CV2KH0</g><l>589</l>, Avg: 0.8591322386490197


java.lang.ArrayIndexOutOfBoundsException: -1
at es.pj.circuits.fpgaft.SimulatorFaultyCircuit.setState(SimulatorFaultyCircuit.java:349)
at es.pj.circuits.control.BISTPIMSeqOnlineFull.evaluate(BISTPIMSeqOnlineFull.java:231)
at es.control.NoisyPIM.evaluate(NoisyPIM.java:133)
at es.pj.circuits.control.CircuitParsimonyPIM.evaluate(CircuitParsimonyPIM.java:80)
at es.control.ShellIM.evaluateNext(ShellIM.java:68)
at es.control.Monica.evolveGeneration(Monica.java:121)
at es.control.Monica.run(Monica.java:255)
at distrit.server.FineGrainedInteractiveTasks$1.run(Unknown Source)
[null, null]


details:

java version "1.4.2_01"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_01-b06)
Java HotSpot(TM) Client VM (build 1.4.2_01-b06, mixed mode)

Linux dmahu 2.4.22 #6 Tue Sep 2 17:43:01 PDT 2003 i686 unknown unknown GNU/Linux
(a slackware 9.1 distro)

$ cat /data/dhep/startIsland
while [ t = t ]
do
java -server -Djava.security.policy=java.policy -jar ITClient.jar 139.184.166.27 KingKong DanielMahu -i 1 -c 2 [my_email_address]
done

the reason I used -i 1 -c 2 was to reproduce the error faster. It originally happened with the default -i and -c

[adream]

all of my clients are doing that today , looks like the serv er is having a bad day, or someone is fiddling with the university network again

i have 12 clients all stalled with similar messages

regards

adream

[DanielMahu]

when the server recoveres, do you think the data would still be successfully transfered? I mean, should I try running the client with longer delays or is it pointless?

[HitchHiker]

More errors: This one started yesterday, evening I think.

Looking for output 2 found as input -1
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
Checking if E High
2004.01.29 13:50:17.717
P0, Gen: 197110, Top: F<f>0.9999999999999922</f><p0>1.0</p0><p1>1.0</p1><p2>0.6785714285714286</p2><g>Q6d2KWfc10oOlRLYNIOvL_qe[hueRk2lRiHL[5qlC6q9rlXK78fhCe_NRAKZtIWrS_7q~1mWIjtV4_5an[YbFo]NKBd^A1u`GTW</g><l>589</l>, Avg: 0.8630257176885278

The part between "looking & the date is repeated some 10 times. It seems to me that some of the circuitdata is faulty.

[adream]

yes same here on nearly all the faster clients......

seems theres a problem... ho hum ... again

[michaelgarvie]

Sorry about all this. That was due to one sneaky little bug. The good thing is that clients will never really crash, they'll stay there waiting for the server to send them new code, which it has now and they're all running again.

Thanks for being so alert! By the way we've just got an even smaller version of the circuit that had beaten sequential hand designed circuits for the first time last week.

[Bok]

miguel,

has this caused the main stats to not update as well. The team scores don't seem to have moved in 48 hours???

Bok

[michaelgarvie]

For starters the client's can't have been crashed for more than 5 hours. At that time those crashed clients would not have been clocking up stats because they weren't doing any useful work. At all other times, and at all times for clients that did not crash, stats were collected as normal. If you check the Daily Rankings you'll see there's was a lot of work logged for yesterday and today.

Basically, stats were not affected by the bug. However see the next message for an explanation of the stats effort being stuck for this couple of days.

[michaelgarvie]

Bok,
Actually you were right. For some reason the effort metric for the current run was 0. Now I've set it to what it should be and all the work logged for this run has been added to the totals. Thanks for pointing this out and sorry for getting you worried!
NB: Nothing has been lost, it was all invisible - multiplied by 0!