PDA

View Full Version : Someone tell me this isn't normal


PCZ
10-15-2003, 05:42 PM
ALL
Yesterday all my lifemapper clients bombed out uploding to the server.
This was definitely server related as all my clients died uploading work
during the same time period.

Now it has just happenend again.
Starting them back up results in an Illegal operation dialog box.


Update

You have to delete any undersize Jobpackets then the client will start again.
Looks like any time the Server is unavailable the client commits suicide.
rshepard
10-15-2003, 06:24 PM
Apparently this is just a problem for the CLI-- their server downtime hit just before I left work,
so I dropped back to the GUI for the evening. Hopefully they'll get this sorted soon.

edit==> Damn, that's quite a performance hit for the GUI - takes ~twice as long to work thru a model

:shocked:
IronBits
10-15-2003, 07:13 PM
When I got home, I saw the cli had bombed out.
Was unable to re-start it. Back to FAH it goes until they fix that problem...
PCZ
10-15-2003, 08:03 PM
IB
You can fix it by deleting the Jobpacket files if they are small 1 or 2k.

I want to step up production a bit on this project but this bug is a show stopper.

Funny edorajh didn't mention this bug when they came here pimpin :mad:
IronBits
10-15-2003, 08:10 PM
I deleted JobPacket01.dat Result01.dat and it took right off!
I run it like lifemapper_cmd.exe -c 10 -p 64
-c caches 10 units, -p 64 runs it at the lowest priority
works great /me tips hat to DP for those options :D
PCZ
10-15-2003, 08:15 PM
IB

Now DP has to get you back in the Stats. I saw you there earlier but you seem to be absent now.
IronBits
10-15-2003, 08:20 PM
I found me here http://www.lifemapper.org/lm-individualmembers/lm-memberstats/StatsMembersOfFree-DCName_1.htm
So I haven't wandered too far. ;)
rshepard
10-15-2003, 08:20 PM
looks like you can't cache more than 20 jobs-- I was trying to cache 48 and it quit at 20
oh well
IronBits
10-15-2003, 08:32 PM
DP told me there are problems if you cache more than 10. :dunno:
rshepard
10-15-2003, 08:40 PM
beats me--- all 4 boxes loaded up 20 as nice as could be, then ran off to start mapping.
10 might be a practical limit if you're on a small pipe or something.
Or, now that I'm back home, they are probably cratering back at the office :trash:
edorajh
10-16-2003, 05:46 AM
Originally posted by PCZ
IB
You can fix it by deleting the Jobpacket files if they are small 1 or 2k.

I want to step up production a bit on this project but this bug is a show stopper.

Funny edorajh didn't mention this bug when they came here pimpin :mad:

:) Sorry PCZ, this bug hit me yesterday for a first time, and only on my box with dial up. My broadband box never has any problem with CLI. :)

Every reported bug could be found here:
http://www.planetspectre.com/board/index.php?showtopic=171


DP told me there are problems if you cache more than 10.

Some mappers report that after Result09, CLI starts to write Result01, and rewriting old Result01. That's why suggestion to use cache not bigger than 9.

Guys from LM team work on new release of CLI that will fix known bugs.

;)
PCZ
10-16-2003, 07:44 AM
edorajh

As for not mentioning the BUG don't worry about it, all the DC Projects have them from time to time.
I hope they sort it soon though before it puts people off the project.
TPR are far too comfortable up there at the top :jester:

Is there a way to run this Project on dialup ?
edorajh
10-16-2003, 07:54 AM
:)

It takes lots of babysitting if you run it with dialup, as it crashes once it finish first jobpacket trying to connect to server. I mainly use CLI on dilaup when I'm online (then everything runs fine). When I'm offline I manualy move Result file to separate folder (after I stoped CLI), and then start CLI again with new JobPacket. When I collect 10-15 Result files I move them back to LM folder and then connect to Net, and start CLI. CLi then send all result files to server.

As I said, lots of babysitting. That's why majority of my output is from my broadband connected box. ;)