AthlonRob wrote:
[...]

The boinc servers are still having some issues; hopefully they'll get

[...]

You can see an explenation here:

http://logs.boinc-irc.com/logs.php?i=2004_06_29

...do a ^F and look for RomWalton.

Nah... I did a good ole' linux grep:

###
* *    RomWalton has joined #Boinc
/12:14:05/    RW:    Howdy
/12:14:49/    RW:    Been busy the last few days...
/12:15:20/    RW:    But things should start getting better now...
/12:15:40/    RW:    Not new per say...
/12:15:57/    RW:    Just brought online some additional machines we
/12:16:42/    RW:    So orginally, BOINC for all three projects was
/12:17:23/    RW:    Now work for the public project is being
/12:17:46/    RW:    We might be adding some more machines to the
/12:18:10/    RW:    in the last 24 hours, an additional 30,000
/12:19:08/    RW:    Well yesterday, we had a backlog of 200,000
/12:20:47/    RW:    Basically on Friday the transitioner wasn't
/12:21:18/    RW:    so when the project started reporting no work
/12:21:55/    RW:    Well that sort of made matters worse since it
/12:22:24/    RW:    instead of just dealing with the existing
/12:22:53/    RW:    So by mid Saturday, we had generated 400,000
/12:23:25/    RW:    with each workunit needing the transitioner to
/12:24:52/    RW:    The four transitioners we have online are still
/12:26:01/    RW:    Well, until i shutdown the apache server we
/12:26:26/    RW:    because incoming results would continue to
/12:27:06/    RW:    with apache down we are chewing through
/12:27:16/    RW:    catch ya laters
/12:29:25/    RW:    Well, right now, the transitioners
/12:29:44/    RW:    but the transitioners are waiting on the db
/12:30:09/    RW:    and apache is stealing to many cpu cycles from
/12:30:32/    RW:    so we changed the scheduler to be a fast-cgi
/12:31:06/    RW:    We are still waiting on the additional memory
/12:31:37/    RW:    2gb for each of the three machines
/12:32:47/    RW:    yeah
/12:35:12/    RW:    i would rather not at the moment...
/12:35:38/    RW:    i would like the transitioners to have full
/12:37:00/    RW:    oh, lol
/12:37:29/    RW:    right now, the list is very small, with no
/12:37:56/    RW:    but the db isn't under very much load at the
/12:38:55/    RW:    When apache is running, the machine load spikes
/12:39:22/    RW:    thats why we migrated it to fast-cgi
/12:39:27/    RW:    as quickly as possible
/12:40:48/    RW:    the ideal solution is to move the web frontend
/12:41:01/    RW:    but we are waiting on the memory sticks for
/12:41:15/    RW:    so, for right now, it has to share
/12:42:05/    RW:    yeah, David and I have been talking a great
/12:42:43/    RW:    also making better use of batching
/12:42:48/    RW:    and transactions
/12:45:25/    RW:    we are in the process of migrating to 1gbit
/12:49:45/    RW:    the frontend is down so the backend can play
/13:11:21/    RW:    :)
/13:11:44/    RW:    Yeah, I brought the whole frontend and stuff
/13:12:00/    RW:    a two day backlog isn't good...
* *    RomWalton has quit IRC
###


Thanks for the (unofficial) details,

Regards,
Martin