Server issues

This entry was published at least two years ago (originally posted on June 26, 2004). Since that time the information may have become outdated or my beliefs may have changed (in general, assume a more open and liberal current viewpoint). A fuller disclaimer is available.

This doesn’t affect this particular site at all, but in case Dad or Kirsten check in, there seem to be issues with the djwudi.com/hanscomfamily.com/geekmuffin.com/interalia.org server at the moment. Phil and I are trying to work on it and find out what the issue is — whatever it is, it popped up sometime during the night, when neither he nor I were doing anything with the machine. No current ETA on when we’ll be back up and running, though we’re trying to bring it back up as soon as possible.

Update: Still down, still no ETA, and I need to get out of the house for a bit. I’ll get more info up when I can.

This also means that any e-mail sent to my [\@michaelhanscom.com]{.citation cites=”michaelhanscom.com”} address will not get to me at the moment. Please use one of my other e-mail addresses if you have one of them and need to get ahold of me.

Just so I don’t lose track or forget about what we tried so far, a quick copy-and-paste of the work so far:

Phil: el problemo, senor

Michael: what’s up?

Phil: i think something has run awry in my crontab
Phil: and is chewing up way too many resources
Phil: on the webserver.

Michael: aah, is that what’s going on
Michael: i just had to restart it – it wasn’t responding to anything when I woke up, except pings

Phil: i suspect you’ll find tons of sa-learn processes

Michael: no ssh or httpd calls would answer
Michael: is it down again now?

Phil: i got ssh to work verrrrrry slowly.
Phil: No responses again now.

Michael: hrm
Michael: well, yeah, something’s borked – the UI is frozen
Michael: gonna restart it again

Phil: gah.
Phil: let me know when you restart it, i’ll ssh in and nuke my crontab before anything launches.

Michael: k
Michael: okay, login window is up
Michael: give it a shot

Phil: i’m in
Phil: gonna watch top and see what launches.

Michael: :nods

Phil: other than a lot of sendmail and procmail processes, nothing seems to be wigging out.

Michael: took out whatever it was that was having an issue? or it just doesn’t seem to be “issuing” right now?

Phil: i took out what i -think- was causing it. no idea if that was it or not.
Phil: but i recall something similar happening last time i tried to make a cron job for sa-learn.

Michael: okay
Michael: what’s sa-learn?

Phil: SpamAssassin’s Bayesian filter learning tool.
Phil: It’s the “learn this now, dammit” tool.

Michael: ah, okay

Phil: i’m still getting some false negatives that it should’ve caught.
Phil: um.
Phil: i don’t think that was it. i’m not getting any responses on ssh again.

Michael: yeah – the UI’s frozen again
Michael: wierd

Phil: goddamnit. what is going on?
Phil: i know i haven’t actually changed anything…

Michael: maybe it’s the machine? (though I hate to suggest that)
Michael: restarting again

Phil: well, we’ll see; i sincerely hope that’s not it.

Michael: you and me both
Michael: (and kirsten and dad too, I’d bet)

Phil: is it acting funny? fans any louder than usual?

Michael: doesn’t seem to be
Michael: i just sits in the corner and hums happily to itself (shrugs)
Michael: as opposed to Marvin, who sits in the corner and hums dismally to himself

Phil: no, i’d wager something is getting overloaded.
Phil: i don’t know what, though. hmmmm.

Michael: could I be getting hit by an attack of some sort, I wonder?

Phil: hmmm.

Michael: god_damn_ that thing takes a while to start up…lol

Phil: there’s a way to test that.

Michael: it’d have to be a pretty good attack to completely lock it up, though

Phil: during the brief time i can get on i’ll try to install snort.

Michael: it’s up

Phil: can’t get it to respond to ssh even now.

Michael: huh

Phil: n/m
Phil: there we go

Michael: okay

Phil: sshd probably wasn’t up yet

Michael: i’m logged in, have top -u 15 running in a term window

Phil: should see a wget process.

Michael: if it locks up, hopefully it’ll give me an idea of what the issue might be (maybe)

Phil: hopefully.

Michael: right now it’s mostly just top at the top of top (whee)
Michael: i’ve seen ssh, gzip, and tar pop up occasionally, but the drop right back down again

Phil: those were me.

Michael: i’ve got my eye on you, young man…lol
Michael: imapd just hit thetop – bouncing between 10 and 25%

Phil: probably mail trying to connect.

Michael: and…freeze
Michael: dammit
Michael: the last thing at the top was just imapd at only 10%

Phil: yeah, i noticed. hmm.
Phil: i’m going to nuke my entire crontab. nothing should be launching, but that doesn’t mean nothing is.

Michael: i’m pretty sure mine’s empty…very sure, in fact
Michael: okies, restarting

Phil: and /etc/crontab just has system stuff in it.

Michael: back up

Phil: yep. i’m in.
Phil: ah HA
Phil: apache2 is starting.

Michael: ?
Michael: i was about to run a permissions check/repair, just for grins and giggles…
Michael: oh, wait

Phil: if you still want to, go for it

Michael: apache2?
Michael: didn’t you kill that?

Phil: yes.
Phil: precisely my point. i thought i did too.

Michael: hmm
Michael: i wonder why it’s starting…and why it hasn’t been a problem until now
Michael: wow

Phil: OK, I need to go find the startup item for apache2 and slay it.

Michael: it’s fixing a lot of permissions

Phil: oh man, wait
Phil: that was NOT a good idea.

Michael: uhoh

Phil: any custom permissions I had set on my website are gonna be borked

Michael: oh shit

Phil: well, we’ll find out soon enough.

Michael: it seemed to be doing a lot in the /System/Library/Perl/ directory
Michael: trying to stop it now
Michael: spinning rainbow cursor at the moment

Phil: frozen again
Phil: from CL too
Phil: so apparently apache2 is not the problem, but it is a problem.

Michael: grrr
Michael: hmmm – the UI isn’t completely frozen
Michael: if I move the mouse, the cursor will jump to another point on the screen after a few seconds
Michael: still, it’s quite unuseable

Phil: probably the permissions check is taking a lot of cpu time.

Michael: gonna reboot again
Michael: christ, this is annoying…lol

Phil: tell me about it.
Phil: i want my email!

Michael: lol
Michael: if all else fails, there’s always the possibility of a reinstall
Michael: which, admittedly, could suck
Michael: system install, getting MT up and running again, making sure the websites don’t break, etc. etc., yadda yadda

Phil: Do an archive & install, then.
Phil: (If you have the disk space.)

Michael: :nods definitely
Michael: i do, there’s still gigs free
Michael: not the way I was planning on spending my saturday, though

Phil: nor I.

Michael: if I do have to do that…think it’s worth putting panther on, rather than jaguar?
Michael: i’d have to figure out the new mail system (postfix instead of sendmail, right?)
Michael: hm

Phil: if only for the fact that you’d have to configure postfix, i’m gonna say no

Michael: ah
Michael: okay
Michael: yeah – “starting apache 2 web server” is part of the startup window messages
Michael: and, we’re up again
Michael: holy shit – we’re locked up again
Michael: okay, the fact that that’s getting faster does not bode well

Phil: i noticed.

Michael: goddammit
Michael: y’know, if it was cloudy, i might not mind as much, but it’s a gorgeous day out there…lol
Michael: up
Michael: according to the apache2 control panel, it’s not running
Michael: even though the startup message was in the startup routine

Phil: ridiculous.

Michael: fuck – locked up
Michael: well, locking up
Michael: in the process
Michael: tried to load the sharing control panel
Michael: spinning rainbow cursor of death

Phil: this is bloody odd.
Phil: something has gone horribly wrong between yesterday afternoon and today.
Phil: i haven’t been around to do anything.

Michael: and i haven’t touched it

Phil: when it starts up, go to the apple menu, check about this mac, and see if anything on the amount of RAM or CPU speed looks unusual.

Michael: up
Michael: 640Mb ram…trying to remember if that’s right

Phil: yes.

Michael: yeah, that’s right

Phil: maxed out for this architecture

Michael: lol
Michael: brainless
Michael: hm
Michael: console log has some conflicts listed

Phil: do tell.

Michael: login window could not find image named ‘bang’

Phil: ….?

Michael: :shrugs
Michael: some startup items failed to launch due to conflicts
Michael: dammit…do I have ichat on that machine? lol

Phil: probably.

Michael: fuck, not that it matters
Michael: i’m losing the UI again
Michael: it just keeps slogging down, slower and slower
Michael: until it doesn’t respond

Phil: i noticed…. so weird.
Phil: incidentally, the apache2 item should now start apache

Michael: lol okay
Michael: it’s working, it’s just responding like XP installed on a 286
Michael: making glaciers look positively snappy, in other words
Michael: i don’t get it
Michael: okay – restarting it, but i’ve unplugged it from the network
Michael: going to see if that makes a difference at all (though I’m not sure why it would)

Phil: we’ll find out.
Phil: if that does it, you’re getting DOS’ed.

Michael: not thinking it’s a DOS or a hack or anything, simply because it’s behind my firewall, so huge amounts of traffic on my line should affect my connectivity/speed on this ‘puter, too

Phil: not if they specifically attacked the webserver by IP address or name.
Phil: well, some slowness, maybe.
Phil: but the webserver would see the majority of it.
Phil: …your windows box isn’t on the network, is it?

Michael: nope
Michael: well, yes, but it’s off
Michael: and on the few times I start it up, the first thing I do is apply any recent security updates
Michael: it’s as up to date as it can be without doing a daily security update check
Michael: okay, this isn’t good

Phil: eh, if it’s off, it doesn’t concern us.
Phil: what?

Michael: the box doesn’t seem to be restarting

Phil: …

Michael: i’m going to unplug it and let it sit for a few minutes

Phil: well, that’s not good.

Michael: given that this started happening when neither of us were doing anything on the box, and the fact that it seems to be a gradual thing – works fine for a bit, then gradually gets slower – and that it seems to be happening sooner on each reboot, and that it doesn’t want to reboot now…
Michael: i’m really starting to worry that it’s a hardware issue

Phil: that’s not good.

Michael: which, realistically, is a possibility – this little G3 was used when I got it, and it’s been running as a server 24/7 for, oh, probably around four or five years now
Michael: the three years since i’ve been in seattle, plus a year or two in anchorage
Michael: it would suck if that’s the case, though
Michael: i don’t have the funds for a replacement
Michael: and i could move things onto the G5, but i’ve liked having a server separate from my work machine

Phil: Yeah, I can definitely understand that… I was gonna ask about that.

Michael: it’s an option, though, should the g3 not come back

Phil: Do you still have the hardware diagnostic CD somewhere for the g3?

Michael: i don’t think I ever had one

Phil: must be a new thing.

Michael: :nods
Michael: either that, or I just didn’t get it with the machine

Phil: well, we have an hour and twelve minutes to figure something out.

Michael: 1:12? lol

Phil: I’m going out to the Irish festival at that point.

Michael: aah
Michael: yeah, I want to get out of the house at some point
Michael: shame to waste a perfectly good kilt day

Phil: amen. if I had a utilikilt i’d wear it today.

Michael: you should do it, man

Michael: not that you’d have it today, but still

Phil: yeah, i know. well, hey, my birthday’s coming up in a few months… we’ll see

Michael: excellent
Michael: okay, booting

Phil: OK, it freezes up, you need to restart in single-user mode and run fsck to see if there are disk errors.
Phil: http://www.macosxhints.com/article.php?story=2004011205473937&query=fsck

Michael: and I do that…how, again? lol

Phil: click the link

Michael: I did
Michael: The specific sequence of how to do this varies depending on what version of OS X you’re running. Go to Apple Support for specific instuctions.

Phil: it’s still detached from the network, right?

Michael: yeah

Phil: OK, let me find something that doesn’t suck i just noticed “check/fix the filesystem”

Michael: lol

Phil: i think you need to hold down S while booting.

Michael: gaaa…doesn’t seem to be booting
Michael: got the startup chime, nothing else

Phil: oh, maaaaan. that’s not good.
Phil: actually, though.
Phil: that might be a sign of a disk error caused by the forced reboots.
Phil: we had the exact same thing happen with a G4 at work.

Michael: :nods it got in last time, and I did a manual (correct) restart

Phil: iMovie froze, the finder crashed and wouldn’t restart, and everything rainbowed.
Phil: I’m going to suggest an archive and install.

Michael: :nods that’s what i’m thinking (grr)
Michael: i’ll deal with that later, though

Phil: just remember to hold down C as you turn it on with the jaguar disk in the driver (or panther; if you want, i can look up postfix guides while you wait)
Phil: k

Michael: right now…i’m a little annoyed, hungry, and want outside lol
Michael: i may do panther…it’s got enough under-the-hood upgrades that it might be a good idea
Michael: just take a little poking around to get postfix up

Phil: and really, from what i recall, setting up postfix wasn’t hard at all.
Phil: i did it at WWDC the day i installed panther.

Michael: as long as the websites survive, i’ll be satisfied

Phil: i think the config files made more sense to human beings

Michael: that’s my big worry, really
Michael: i’d hate to end up nuking everyone’s sites (again)

Phil: i do too. y’know, if all else fails, stick the HD into the G5. I assume it has an expansion slot for a second HD.

Michael: yeah, but I can’t just drop the drive in
Michael: different drive specs
Michael: serial ATA on the G5

Phil: oh, fuck, that’s right. dammit.

Michael: i’d need to drop the old drive(s) into ATA cases

Phil: if nothing else, even if the machine won’t boot, our data is still there.

Michael: yeah

Phil: you’re good, webmaster. go get food and sunshine; i’ll be doing the same myself shortly.

Michael: fuck – and i’m using michaelhanscom.com for most of my e-mail right now, too…lol
Michael: argh
Michael: ah, well
Michael: it’ll wait for a day
Michael: or a few hours, or whatever
Michael: anyway…i’m wandering – enjoy the irishfest

Phil: will do. later man

Update: Everything’s back up again. Phil tracked it down to a cron job that had piled a few thousand e-mail messages into his inbox, which caused everything to choke. Inbox is empty, cron is smacked back into submission, and all’s well that ends well.

iTunes: “Three Wishes” by Waters, Roger from the album Amused to Death (1992, 6:50).