Welcome to the MacNN Forums.

Shaktai · Apr 20, 2004, 01:12 AM

I haven't been able to access the team stats for a few hours now, for any project. Tried from a couple of different computers just to be sure it wasn't a local problem. Anyone else having the same problems?

billybob128 · Apr 20, 2004, 02:36 AM

i cant connect either just times out on me before it goes through

Scotttheking · Apr 20, 2004, 05:11 AM

It's a rule, stats will die if I'm out of town.

*Scott notes that it's 2am, he has to make a speech in 7 hours, and has to write the speech first*

chris v · Apr 20, 2004, 09:17 AM

Originally posted by Scotttheking:
It's a rule, stats will die if I'm out of town.

*Scott notes that it's 2am, he has to make a speech in 7 hours, and has to write the speech first*

Come on. Get your priorities straight and fix the server.

CV

Shaktai · Apr 20, 2004, 10:54 AM

Originally posted by Scotttheking:
It's a rule, stats will die if I'm out of town.

Forgot about that rule.

Okay, today is the 20th, I guess we can live for a few more hours without stats.

reader50 · Apr 20, 2004, 02:14 PM

The current setup is box#1 as the webserver and box#2 as the DB server and stats scripts box.

Box#2 can't be reached, it's not obvious if it has crashed or is running a super load.

I've turned off the stats crons for box#2, if it's a super load it should recover within an hour. The last time this happened, Scott was away also - we rebooted but never found out why box2 had bogged down, nothing obvious was in the logs.

If Scott will be back soon or can otherwise reach box#2 online, I'd like to avoid a reboot and let him look it over, see if we can figure out what caused the problem. Perhaps some application has a memory leak and it's dying under VM swap.

If he can't look it over soon, we'll have to reboot, do repairs, and continue.

reader50 · Apr 21, 2004, 02:21 AM

We still don't know what caused box#2 to become unresponsive, but it's been rebooted and the DB has been repaired. Stats are back up.

The computer had not crashed, and allowed a local login - but nothing happened after it accepted the login. No obvious HD activity, which seems to rule out VM swap death. The fact that it wasn't crashed rules out most hardware issues, with the possible exception of HD related problems such as connector cables.

Checking the SMART results might reveal something, the system logfiles had nothing unusual - which would be consistent with a HD issue. No logs would be written after the HD issue hit.

My best guess is an intermittent issue related to the hard drives, such as an overheating RAID card - but both failures happened either late at night or fairly early in the morning, not the likeliest times for overheating. Scott will have to do some detective work to narrow it down any further. The last time stats went down, the symptoms were the same, so this is a recurring issue that we should solve.

If anyone has evidence that the aliens are behind this because we're neglecting SETI, I would be interested.

Welnic · Apr 21, 2004, 03:15 AM

I think that it would be the other way around, they would not want us to support SETI. If they wanted to be found they could just show up. If they want to hide then they would do something like cause BOINC to be in endless beta hell.

Scotttheking · 02:01 AM

I'd love to replace the pair of servers with a single server, but it'd run as much as $4K to get a single box to handle teamstats, and handle it well.
That'd be a dual opteron box in the 1.6 or 1.8GHz range, with 2GB of ram (the more the better, 8GB would mean the whole DB could live in ram), 2 74GB WD Raptor SATA drives in raid1, and a pair of 250GB drives in raid1 for DB backups, a 1U rackmount case, and some other niceties. If I could afford it I'd go for it.

I haven't figured out what's wrong with the database server. It's died twice, both when I'm out of town.
Both times it wouldn't let me login, either remotely or locally.
Both times the logs showed nothing out of the ordinary.
It'll be something to worry about after school gets out in a few weeks.

Edit: FYI, the support the team stats thread has some affiliate links in it, and the team front page has a google ad. That's what I'm using to fund server changes.

osiris · Apr 29, 2004, 10:03 AM

Did Scotttheking go out of town again?

I think the stats page is down, no update in 14 hours.

Scotttheking · 12:54 PM

you've got to be more specific then "stats are down". "The seti stats haven't updated in 16 hours" is more useful.
And stats are fine, it's probably their end that isn't working.

osiris · 01:21 PM

Originally posted by Scotttheking:
you've got to be more specific then "stats are down". "The seti stats haven't updated in 16 hours is more useful."
And stats are fine, it's probably their end that isn't working.

Sorry for the inconvenience.

Anyway, the SETI stats page is updating now. Thanks